Stephen Priest

Page 1

05/12/2007

KANTS CONCEPT OF FREEDOM in the Critique of Pure Reason

Stephen Priest

Stephen Priest is a member of the Faculty of Philosophy in the University of Oxford. He is Senior Research Fellow of Blackfriars Hall, Oxford and a member of Wolfson College, Oxford and Hughes Hall, Cambridge. He is the author of The British Empiricists, Theories of the Mind, Merleau-Ponty, and The Subject in Question, editor of Hegels Critique of Kant, Jean-Paul Sartre: Basic Writings and coeditor (with Antony Flew) of A Dictionary of Philosophy. Stephen Priest has lectured widely in the United States and Europe and his writing has been translated into Dutch, Spanish, Russian, Japanese, Korean and Macedonian.

Stephen Priest December 2007

Stephen Priest

Page 2

05/12/2007

Preface In the Critique of Pure Reason, especially in the Third Antinomy, Kant offers arguments for a solution to the problem of freedom and necessity which seem to do justice to the claims of libertarianism and determinism. My restricted aim in this short book is to decide whether any of them is sound. Because this is a philosophy book, not a work in the history of ideas, I have not attempted to reconstruct the whole of Kants thinking about the self, nor to locate it in a historical context. In philosophy, the relevant units are the problems, their putative solutions and the arguments for those putative solutions. My approach to Kant is therefore more like Jonathan Bennetts than those of Karl Ameriks or Paul Guyer. By understanding Kant I mean the allocation of propositions to his sentences in a way that maximises their consistency as a set. The justification of this method is: To the extent to which some sentences form an inconsistent set they express nothing. I point out places where Kants arguments have been misunderstood and substitute interpretations I consider more viable. On the key issue of the viability of Kants compatibilism I adjudicate between the major commentators. It is a Leitmotif of the literature that Kants theory of freedom is a failure. In Kant Ralph Walker says Unfortunately Kants attempt to reconcile noumenal freedom with the thoroughgoing determinism of the phenomenal world is a hopeless failure, as has often been pointed out (Walker, 1978, 148) and Walker concludes his account of freedom is not a success. (Walker, 1978, 150) I try to provide Kants theory with more plausibility than it has been granted, partly by interpreting it in a way that exhibits its internal consistency. I am grateful to A. J. Ayer, Graham Bird, Edward Craig, Daniel Came, Benedikt Gcke, Peter Hacker, Michael Inwood, Adrian Moore, Terence Penelhum, Bernard Williams and Timothy Williamson for discussion of Kants philosophy or problems bearing on freedom and determinism or both. I am grateful to the Philosophische Fakultt of the Westflische WilhelmsUniversitt Mnster, Germany for their hospitality in 2006. In particular I thank, Prof. Dr. Andreas Htteman, Prof. Dr. Oliver Scholz and Prof. Dr. Peter Rohs for useful discussion of the invited lectures I delivered there about Kant. I also thank the Faculteit der Filosophie of the Radboud Universiteit, Nijmegen, Netherlands, for their hospitality, also in 2006. In particular I thank Professor Graham Lock, and Tjeerd and Sjoerd van Hoorn for their contributions to a two-day seminar generously devoted to my ideas. Stephen Priest Oxford Michaelmas Term 2007

Stephen Priest

Page 3

05/12/2007

Contents I. Kant on Uncaused Causes....4 The Third Antinomy... 5 Thesis ..5 Antithesis ....5 An Ambiguity .....5 Proof of the Thesis ..7 Kant's Observation on the Thesis ....7 Individuating Causal Chains..12 Proof of the Antithesis 13 The Argument from the Unity of Experience....13 The Empirical Argument...15 The Rules Argument..17 II. Kants Compatibilism...18 Practical Freedom..19 Transcendental and Practical Freedom..22 (a)The Argument for Practical Freedom....22 (b) The Argument from Practical to Transcendental Freedom.....23 My Freedom....29 Allison, Guyer, Irwin......33 Freedom and Self-Knowledge....38 Freedom and Pure Apperception38 Bennett on the Third Antinomy..40 III. Kant on the Freedom of Reason...45 Reason and Freedom..49 (a) The Argument for the Freedom of Reason...49 (b) The Argument that Reason is Not Determined.52 Notes Bibliography

Stephen Priest

Page 4

05/12/2007

I Kant on Uncaused Causes I clarify the role of the concept of an uncaused cause in Kant's putative solution to the problem of free will and determinism in the Antinomies chapter of The Critique of Pure Reason and offer some comments. Kant's most concise yet clearest definition of 'freedom' is the statement that freedom is 'self-activity' (Selbstttigkeit) (A417, B446). I shall understand some action or event, 'e', to be free on this account if and only if e is either the cause of itself or the cause of some further action or event e ' but is not itself the effect of any cause. So, on either construal, e has an effect, but on the first its effect is not other than itself, but on the second its effect is other than itself. On the first construal, the event has a cause but that cause is only itself but on the second construal, the event has no cause. By 'effect' I mean 'caused event' and by 'cause' I mean 'event with an effect'. (I assume that actions are a species of event.) This is a sharpening of Kant's claim (A417, B446-7) that 'unconditioned causality in the [field of] appearance is called freedom'. [die unbedingte Kausalitt [] in der Erscheinung [heit] die Freiheit [] ]. (Kemp Smiths field of translates nothing in the German here.) Kant says 'the condition of that which happens is entitled the cause'. [Da heit nun die Bedingung von dem, was geschieht, die Ursache.] Kant does not say what sort of condition he has in mind. He might mean necessary condition such that, if e is a condition of e (NC) -c -e or he might mean sufficient condition such that: (SC) c e Kants use of the in the condition does not logically guarantee which condition. He might be using the in a generic sense roughly equivalent to any so that (NCU) (c, e) -c -e or (SCU) (c, e) c e

Stephen Priest

Page 5

05/12/2007

I shall read 'condition' here to mean 'both necessary and sufficient condition'. So, some event e is a caused event if and only if there exists or existed at least one other event c such that if c failed to occur then e could not occur, and if c does occur then e could not fail to occur: (N&SC) (c, e) ((-c - e) & (c e)) Kant's discussions of freedom and determinism presuppose an ontology of events. This is implied, for example, by his use of 'that which happens' (was geschieht) (for example at A417, B446). I shall follow Kant in this, but it obviously needs to be born in mind that some condition for an event's occurrence might not itself be an event. Kant thinks 'I am free' is true only if there is at least one respect in which I am the cause of some event or events, and at least one respect in which I am not the effect of any event. This is so far a necessary but not a sufficient condition of my freedom because my being the cause of some event might itself be caused, if the respect in which (putatively) I am uncaused does not preclude this. So, I am free if and only if the respect in which I am the cause of an event is just the respect in which I am uncaused. This precludes my being caused to be the cause of at least one of the events of which I am the cause. On the other hand, I am not free but wholly determined if and only if even if I am the cause of at least one event, I am caused to be the cause of any event of which I am the cause. With these definitions, and criteria for analyzing free in mind we may turn to the third antinomy. The Third Antinomy I state the thesis and antithesis of the third antinomy and then examine the proof of each: Thesis 'Causality in accordance with laws of nature is not the only causality from which the appearances of the world can one and all be derived. To explain these appearances it is necessary to assume that there is also another causality, that of freedom' [Die Kausalitt nach Gesetzen der Natur ist nicht die einzige, aus welcher die Erscheinungen der Welt insgesamt abgeleitet werden knnen. Es ist noch eine Kausalitt durch Freiheit zu Erklrung derselben anzunehmen notwendig.] (CPR 409, A445, B473) Antithesis

Stephen Priest

Page 6

05/12/2007

'There is no freedom; everything in the world takes place solely in accordance with the laws of nature.' [Es ist keine Freiheit, sondern alles in der Welt geschieht lediglich nach Gesetzen der Natur.] (CPR 409, A445, B473). An Ambiguity In understanding both proofs an important ambiguity needs to be dispelled. Kant is appraising the claim that there exists a kind of freedom from which 'the world of appearances may be derived'. [die Erscheinungen der Welt abgeleitet werden knnen] This could mean either an uncaused event which was the initiator of every subsequent causal chain which existed, or more narrowly, an uncaused event which was the initiator of some single causal chain which is just one member of the set of all causal chains. The first sort of freedom putatively pertains to God, or the cause of what exists. The second sort putatively pertains to finite persons, or performers of causally efficacious actions within the world. In what follows I shall confine the discussion entirely to the question of the freedom of (human) persons and ignore any passages which can only be given the first interpretation. This is because we are only concerned with the freedom of the self here and this is Kant's own expressed concern in writing the Third Antinomy. At CPR 414, A450, B478, he explains: '[[...]] the absolute first beginning of which we are here speaking is not a beginning in time but in causality. If, for instance, I at this moment arise from this chair, in complete freedom, without being necessarily determined thereto by the influence of natural laws, a new series, with all its natural consequences, in infinitum, has its absolute beginning in this event, although as regards time this event is only the continuation of a preceding series' [Denn wir reden hier nicht vom absoluten ersten Anfange der Zeit nach, sondern der Kausalitt nach. Wenn ich jetzt (zum Beispiel) vllig frei, und ohne den notwendig bestimmenden Einflu der Naturursachen, von meinem Stuhle aufstehe, so fngt in dieser Begebenheit, samt deren natrlichen Folgen ins Unendliche, eine neue Reihe schlechthin an, obgleich der Zeit nach diese Begebenheit nur die Fortsetzung einer vorhergehenden Reihe ist.] Kant means that if an event were the cause of every subsequent event then that event would necessarily also be the first of those events in time. This is to be contrasted with free actions performed by a finite person, such as Kant himself rising from his chair. Although, if free, such an action is according to Kant's theory a cause of subsequent events but not itself caused, it differs from the cause of all events in (at least) being temporally preceded by earlier events (none of which is ex hypothesi the

Stephen Priest

Page 7

05/12/2007

cause of that putatively free action). This is not just incidental to the passage because Kant thinks there is a strong interdependence between being free and being able to use 'I'. Proof of the Thesis The proof takes the form of reductio ad absurdum. Kant assumes the contradictory of what is to be proved and attempts to derive a contradiction. On the assumption that any event is causally necessitated by some temporally precedent event, (including a fortiori any event which precedes and causes any event), so there will be 'no completeness of the series' [keine Vollstndigkeit der Reihe] (CPR 410, A446, B474) on the side of the causes that arise one from another' [auf der Seite der von einander abstammenden Ursachen]. In other words, determinism entails an infinite regress of caused events. This according to Kant is 'self-contradictory'. Kant does not make explicit what the contradiction is but I think the following can be reconstructed: There both was and was not a first event if (per impossibile) complete determinism were true. This is because any putative first event would have to turn out to be caused by some temporally prior event. But such a series could never have begun if there had not been just one event of which every subsequent event were a causal consequence, but to claim that conjunction is precisely to require what is logically impossible (or self-contradictory to describe): that there both was and was not an uncaused first event. This argument for the thesis of the First Antinomy leaves room for human freedom but does not entail it. Kant thinks the way out of the regress is to halt it. The way to escape from the contradiction is to deny that there is complete determinism: 'We must, then, assume a causality through which something takes place, the cause of which is not itself determined in accordance with necessary laws, by another cause antecedent to it, that is to say, an absolute spontaneity. [Diesemnach mu eine Kausalitt angenommen werden, durch welche etwas geschieht, ohne da die Ursache davon noch weiter, durch eine andere vorhergehende Ursache, nach notwendigen Gesetzen bestimmt sei, d.i. eine absolute Spontanitt.] (A446, B474-5) I shall understand an action or an event's being an 'absolute spontaneity' to mean its being a cause or initiator of a causal chain, but its not itself being an effect of any cause or causal chain. So, e is an absolute spontaneity, if and only if there exists at least one event e' temporally subsequent to e of which the occurrence of e is both a necessary and a sufficient condition, but there exists no event e-1 temporally anterior to e such that the occurrence of e-1 is either a necessary or a sufficient condition (or both) for the occurrence of e. (1)

Stephen Priest

Page 8

05/12/2007

From the fact that there must be postulated an absolute spontaneity to make some sense of the origin of what exists, it does not follow that part of what exists is free action performed by human persons even if human action is itself well described as an absolute spontaneity. So the argument for the thesis of the third antinomy does not amount to a proof of 'I am free'. Kant's Observation on the Thesis Kant admits the thesis fails to establish libertarianism as a truth about persons when he says: 'The necessity of a first beginning due to freedom, of a series of appearances we have demonstrated only in so far as it is required to make an origin of the world conceivable' [Nun haben wir diese Notwendigkeit eines ersten Anfangs einer Reihe von Erscheinungen aus Freiheit, zwar nur eigentlich in so fern dargetan, als zur Begrifflichkeit eines Ursprungs der Welt erforderlich ist.] (A449, B476) implying that further argument is required to show that 'free' may be truly predicated of human actions. Does Kant have such an argument at (A448, B476-A52, B480)? Kant analyses the concept of freedom into two sub-concepts; transcendental freedom and empirical freedom, and calls the whole concept thus sub-divided 'psychological freedom'. (A448, B476) He has already explained what transcendental freedom is. (A446, B474). It is a kind of uncaused cause 'without which even in the (ordinary) course of nature, the series of appearances on the side of the causes can never be complete'. This already introduces the shift in his concern from an uncaused origin of what exists to the freedom of the human agent. Otherwise he would have omitted 'appearance' and simply written 'series of causes' or 'series of events'. So, putatively, there is a kind of freedom which is a condition of appearances in particular not merely of events in general. I take it that any transcendental cause of appearances will not itself be an appearance because on Kant's view something's being truly characterised as transcendental precludes it's being truly characterised as empirical, and appearances are empirical. Still, he says the content of the psychological concept of freedom is 'mainly empirical' [welcher groen Teils empirisch ist.] (A448, B476) I shall interpret this to mean that, more often than not, the predicate 'free' and the noun 'freedom' feature in sentences which may be verified and falsified empirically. But clearly, on Kant's account that is not the only use of the two terms. The idea of transcendental freedom does not have any empirical use, yet must be invoked if human action is to be correctly described as free: 'The transcendental idea stands only for the absolute spontaneity of an action, as the proper ground of its imputability.'

Stephen Priest

Page 9

05/12/2007

[Die transzendentale Idee der Freiheit macht zwar bei weitem nicht den ganzen Inhalt des psychologischen Begriffs dieses Namens aus [] sondern nur den der absoluten Spontaneitt der Handlung, als den eigentlichen Grund der Imputabilitt derselben.] (A448, B476) Indeed, 'imputability' here implies that a certain sub-class of events could not be truly described as actions unless they were transcendentally free. It clearly follows that transcendental freedom is a condition of empirical freedom because unless actions could be attributed to persons there could not exist true empirical sentences about those persons and their actions implying their freedom. Although this is the position in the observation on the thesis of the third antinomy Kant is clearly unhappy about it. He says; This, however, is for philosophy a real stumbling block; for there are insurmountable difficulties in the way of admitting any such unconditioned causality.' [ist aber dennoch der eigentliche Stein des Anstoes fr die Philosophie, welche unberwindliche Schwierigkeiten findet, dergleichen Art von unbedingter Kausalitt einzurumen. (A448, B476) For example, Kant finds it difficult to make sense of the idea of 'a power of spontaneity beginning a series of successive things or states' [Kemp Smiths spontaneity translates nothing in the German in this passage. Kant has at B476: ein[em] Vermgen [], eine Reihe von sukzessiven Dingen oder Zustnden an[zunehmen (original; angenommen werden msse)], that is, exactly a cause that is not an effect. But, on the argument so far, that is required if the will is free. Kant's solution is to side-step what is in fact a genuine philosophical problem. He simply asserts that the idea of a caused event is just as obscure as that of an uncaused event, yet (because 'cause' is a category) we have 'a priori knowledge that this latter type of causality must be presupposed' (A448, B476) [da wir uns eben sowohl bei der Kausalitt nach Naturgesetzen damit begngen mssen, a priori zu erkennen, da eine solche vorausgesetzt werden msse]. In other words, we are compelled to think of our experience as of causally related events as one of the conditions of our finding it intelligible. The only genuine question now remaining on Kant's view is whether we may legitimately postulate uncaused causes to similarly make intelligible our empirical belief that we are agents who perform actions freely. Thus he concludes: 'How such a power is possible is not a question which requires to be answered in this case, any more than in regard to causality in accordance with the laws of nature.'

Stephen Priest

Page 10

05/12/2007

[Wie ein solches mglich sei, ist nicht eben so notwendig beantworten zu knnen, da wir uns eben sowohl bei der Kausalitt nach Naturgesetzten damit begngen mssen.] (A448, B476) So, 'Am I free?' construed as a transcendental question is given no answer in the observation on the thesis of the third antinomy. Before turning to Kant's substitute question it should be noted that the concept of an uncaused event is not inherently incoherent. From the fact that some event occurs it does not follow that it was caused to occur. Nothing turns on accepting these as examples but, to illustrate the logic, suppose an event was the only event that ever happened or, think of everything as an event. It is at least not self-contradictory to assert that an event occurs but is not thus caused to occur. Kant argues that we may coherently postulate uncaused causes so that we may think of our wills as free, and his argument takes the form of removing a misunderstanding, and pointing out a parallel with natural determinism. I'll take the parallel with determinism first. The first premise is 'We are not in the least able to comprehend how it can be possible that through one existence the existence of another is determined.' [Ob wir gleich die Mglichkeit, wie durch ein gewisses Dasein das Dasein eines andern gesetzt werde, auf keine Weise begreifen.] (A448, B476) The second premise is 'But the power of spontaneously beginning a series in time is thereby proved (though not understood)' [Weil aber dadurch doch einmal das Vermgen, eine Reihe in der Zeit ganz von selbst anzufangen, bewiesen (obzwar nicht eingesehen) ist.] (Here we assume the reconstructed piece of argument from page 2; otherwise the causal chain could not begin) and the conclusion is: 'It is now also permissible for us to admit within the course of the world different series as capable in their causality of beginning of themselves, and so to attribute to their substances a power of acting from freedom' [so ist es uns nunmehr auch erlaubt, mitten im Laufe der Welt verschiedene Reihen, der Kausalitt nach, von selbst anfangen zu lassen, und den Substanzen derselben ein Vermgen beizulegen, aus Freiheit zu handeln.]

Stephen Priest

Page 11

05/12/2007

There are two strands of thinking here. One is; we are compelled to think of the world as containing deterministic causal chains, even though the precise nature of causal connection is obscure to us. Therefore we are entitled equally to think the world as containing undetermined events which are initiators but not effects of causal chains even though the notion of 'free' or uncaused causes may be obscure to us. The second strand is that determinism is only thinkable if incomplete. In particular, if there is a deterministic chain then it must have had a beginning, and so a first event which initiated all subsequent events but was not itself initiated by any prior event. But this contains the idea of uncaused cause that is needed for us to think of our own actions as free. So, determinism alone does not preclude freedom of the will, or my rightly answering 'yes' to 'I am free?' even when this is a transcendental question. Clearly though, this does not prove 'I am free'. I turn next to the misunderstanding that Kant warns us against. This is the assimilation of the start of a causal chain to the start of a mere chronological series. If we fail to draw this distinction we will be led to erroneously suppose an uncaused cause impossible unless it is the first cause. Someone who failed to draw the distinction might argue that because an allegedly free action would not be a first cause but would be only the putative initiator of a causal chain it would be 'always preceded in the world by some other state of thinks' (448, B476) [indem doch immer ein Zustand der Dinge in der Welt vorhergeht] and so would not be itself uncaused. Kant is clearly correct to argue against this way of thinking because, even in the case of a caused event, not every event preceding it in time is a cause of it even if some are. So from the fact that some event occurs, and its occurrence is temporally preceded by earlier events, it does not follow that it is caused by any one of those earlier events. This removes one thinkable line of objection to the truth of 'I am free'. (2) Individuating Causal Chains Kant draws a distinction between an absolute and a relative beginning (A451, B476) and wishes to show the possibility that the uncaused causes necessary for human freedom may be absolute and not merely relative beginnings. I shall read the distinction this way: An event is an absolute beginning if it is an initiator of a causal chain but is not itself an effect. An event is a relative beginning if it is the initiator of a causal chain but is temporally preceded by at least one earlier event. Kant wishes to leave room for an event to be both an absolute and a relative beginning; that is, an event that is a cause but not an effect and yet which is temporally subsequent to earlier events (none of which are its cause). If human actions are free, then they will turn out to such events. Kant's example 'I at this moment arise from this chair' (A451, B479) is designed to be an example of just such an event. Kant claims that if he arises from his chair

Stephen Priest

Page 12

05/12/2007

'in complete freedom, without being necessarily determined thereto by the influence of natural causes, a new series, with all its natural consequences in infinitum, has its absolute beginning in this event' [Wenn ich jetzt, vllig frei, und ohne den notwendig bestimmenden Einflu der Naturursachen, von meinem Stuhle aufstehe, so fngt in dieser Begebenheit, samt deren natrlichen Folgen ins Unendliche, eine neue Reihe schlechthin an.] There is a specific criterion for the individuation of event series at work here. It is this causal or genetic one: If two putatively numerically distinct event series S and S' were in fact causally initiated by one and the same event then under at least one description S and S' are numerically the same event series. If two putatively numerically identical event series S and S' were in fact causally initiated by numerically distinct events then S and S' are in fact numerically distinct event series. (The second half of the criterion has more plausibility than the first). As Kant deploys this criterion in his example, an event series is initiated by an event and that event has no cause, but every event series does have some event as its initiating cause. The 'new series' exists under a different description from the 'natural' event series. Clearly this is consistent with his view that if I freely rise from my chair then that action may be an absolute and not merely a relative beginning because presumably some events preceded my rising from my chair (or any other action I perform), even though possibly my action qua that is uncaused. There then follows a claim that may be false: '[[...]] this resolution and act of mine do not form part of the natural succession of purely natural effects, and are not a mere continuation of them. In respect of its happening, natural causes exercise over it no determining influence whatsoever.' [Denn diese Entschlieung und Tat liegt gar nicht in der Abfolge bloer Naturwirkungen, und ist nicht eine bloe Fortsetzung derselben, sondern die bestimmenden Naturursachenhren oberhalb derselben, in Ansehung deser Erugnis, ganz auf.] (CPR 414, A451, B479) From the fact that if I am free (in for example rising from my chair) then I initiate a new causal chain, it may indeed be validly inferred that my decision and action are not causally determined. This follows logically given the sense Kant has allocated to 'free'. But the validity of an inference is no guarantee of the true of its conclusion. For that the truth of the premises is also required, and Kant has not demonstrated the truth of 'I am free' amongst the premises. For Kant the postulation of an uncaused cause is a 'requirement of reason'. (A451, B479) We have to make use of the notion to make sense of human decision and action, and of the origin of the universe. But from the fact that we seem

Stephen Priest

Page 13

05/12/2007

psychologically compelled to believe that p it does not follow that p. The fact that much of my thought and action would be unintelligible to me if I did not at least assume the truth of 'I am free' is itself no guarantee of the truth of that assumption. Nevertheless, that the assumption is true is a plausible explanation of its being psychologically compelling. At least, if it is held to be false then some account is needed of why it appears so evidently true. (3) Proof of the Antithesis If the argument of the thesis does not prove 'I am free' does the argument of the antithesis disprove it? It is in fact the conjunction of three arguments. If any one of them goes through then 'I am free' is false in the sense of 'freedom' defined at (A417B446). I shall call these: The Argument from the Unity of Experience, The Empirical Argument, and the Rules Argument. Each has as its conclusion the impossibility of transcendental freedom. The Argument From the Unity of Experience The first premise is (1) 'Transcendental freedom [[...]] renders all unity of experience impossible.' [Also ist die transzendentale Freiheit [] eine solche Verbindung [] nach welcher keinerlei Einheit der Erfahrung mglich ist.] (A447, B475) The second is (2) [Transcendental freedom] abrogates those rules though which alone a completely coherent experience is possible.' [Transzendentale Freiheit [] [reit] den Leitfaden der Regel [ab], an welchem allein eine durchgngige zusammenhngende Erfahrung mglich ist] (A447, B475) An unstated assumption is (3) But there is rule governed unity of experience and the conclusion is in the antithesis of the third antinomy. and the conclusion is 'There is no [transcendental] freedom.' [Es gibt keine [transzendentale] Freiheit]

Stephen Priest

Page 14

05/12/2007

(A445, B473) The form of the argument is (p -q) (q -p), which is a valid pattern of reasoning. But are the premises true? Let us accept (3) in the sense in which it is accepted by Kant. Then the conjunction of (1) and (2) amounts to the claim that there could be no unity of experience if there were uncaused events. In particular, we are interested in whether if I and other persons act freely, experience thereby must be chaotic to us, but two preliminary points need to be dealt with. Arguably if all or most events were uncaused then our experience might be so unpredictable as to be unintelligible and chaotic. But a quasi-Humean reply could be that our experience could exhibit a regularity that was not causal where events of type e were invariably subsequent to events of type e but es were not causes of es. Secondly, if there were an uncaused past event which was the cause of all subsequent events this need not render our present experience of those events unintelligible. But suppose the claim is restricted to the case where if a person acts freely their behaviour would be so unpredictable as to be unintelligible to us. This would perhaps introduce a strong arbitariness into attempts to understand others through our experience of them. There are two replies to this. Firstly, from the fact that I am free (ie that my actions and decisions are uncaused causes) it does not follow that other persons cannot predict my behaviour. This is because I might exercise my freedom with a regularity with which other persons might become acquainted, and they might base their predictions on that acquaintance. Secondly, there is notoriously a difficulty in predicting human behaviour with the certainty that is possible in the study of certain natural occurrences. This measure of unpredictability is on the one hand not so severe as to 'render all unity of experience impossible' and indeed may be due precisely to the exercise of freedom in Kant's sense. So, the Argument From the Unity of Experience, although valid, is not sound because one of its premises is false. Clearly then, it cannot be concluded from it that there is no transcendental freedom. Suppose though we read the Argument From the Unity of Experience in a still more restricted form, so that it is this claim: If all my decisions and actions were free then all my experience would become intelligible to me, but my experience is not intelligible to me, so I am not free. Kant leaves the proponent of transcendental freedom open to this kind of objection because he does not explain the relationship between decisions and actions. But the account could be supplemented on Davidsonian lines to protect it. It is necessary to claim that my decisions' being free decisions consists in their being uncaused causes could be retained, and it could be argued in addition that I might find my decisions predictable by being acquainted with which decisions I had made previously. It is true in any case that I do not often know what I will decide before I decide it. Indeed, knowing what I will decide (because of what 'knowledge' means) could be taken to entail deciding anyway. I call the supplement to Kant's theory of freedom 'Davidsonian' because it allows decisions to be described as causes of actions. This strategy requires sacrificing the

Stephen Priest

Page 15

05/12/2007

transcendental freedom of action because actions are not uncaused causes if they are caused by decisions, but this only amounts to a weakening of the claim 'I am free' because it leaves room for me to be, by my decisions, the uncaused cause of those actions. And complete determinism - the 'there is no freedom' of the antithesis requires the falsity of even this weakened claim. So, even when re-written in first person singular form the Argument From the Unity of Experience is not a refutation of libertarianism. The Empirical Argument The premises of the Empirical Argument are (1) '[Transcendental freedom] is not to be met with in any experience; [Also ist die Transzendentale Freiheit [] [so dass sie] also auch in keiner Erfahrung angetroffen wird] (A447, B475) and (2) [Transcendental freedom] is therefore an empty thought entity' [[Transzendentale Freiheit ist] mithin ein leeres Gedankending] and the conclusion, as before, is There is no [transcendental] freedom. [Es gibt keine [transzendentale] Freiheit] This argument requires as an assumption the version of concept-empiricism which I have called elsewhere the Sense and Sense Argument. According to this doctrine a concept has meaning if and only if it admits of empirical (ie ostensive) definition or else it may be completely translated into a concept or concept that admits of empirical (ostensive) definition. Without this assumption there is no reason why if transcendental freedom is not met with in any experience it should be 'an empty thought entity'. I understand by 'an empty thought entity' a putatively conceptual item which has no existing reference and might have no informative use. (A Gedankending is just an object of thought, but prima facie, according to the German use of the term, it might be informative. It might be part of a Gedankenspiel. In this sense, Pegasus might be a Gedankending.) Premise 1 is analytic; it is part of the meaning of 'transcendental' that if something is transcendental it cannot be an object of experience. The inference from (1) to (2) requires the truth of concept empiricism; then the conclusion follows

Stephen Priest

Page 16

05/12/2007

directly from (1) and (2) because if 'empty' means 'without any referent' and if 'freedom' is an empty thought entity, then 'freedom' has no referent which is equivalent to saying there is no freedom. The doctrine of transcendental freedom is vulnerable to this argument in a way that the doctrine of empirical freedom is not. Although 'free' has empirical uses in sentences whose truth conditions are 'to be met with in experience' when we commonsensically talk of someone being free or not free to do such and such, 'transcendentally free' has no such uses. Only two strategies seem plausible in resisting this line of argument. One is to weaken the concept of transcendental freedom, the other is to find reasons for denying concept-empiricism. Weakening 'transcendental freedom' is difficult just because 1 is analytic. However, it is sometimes argued by libertarians that we are each directly acquainted with our own freedom of will as a kind of experiential datum, that is qualitatively distinct from our experience of series of events in the external world. Although clearly this is to collapse transcendental freedom into empirical freedom and so not to save transcendental freedom from the first premise. A more promising version might be this: we are each, at least sometimes, directly acquainted with our own decisions as or soon after they come to mind but, at least sometimes, we are ignorant of any cause which precipitated our deciding one way or the other. We allow also that our decisions are (amongst the) causes of our actions. If so, then there exists an event of which the following is true: it is a cause but at least is not known to be an effect. This leaves room for the possibility that such an event is in fact a cause that is not an effect. This is because there are two reasons why it might appear to be a cause that is not an effect, viz either it is false that it is a cause that is not an effect but we are not acquainted with the cause of it and so do not know that it is an effect, or else, it is in fact a cause that is not an effect and that is the reason we are not acquainted with any cause of it. This argument leaves room for the transcendental freedom of decisions (but not actions). The premise that we are introspectively aware of what we are deciding or have decided is not a threat to transcendental freedom, because even if what one is thus aware of is, in fact a transcendentally free event, no-one is thus aware of it qua transcendentally free event, which is impossible in Kantian terms: nothing transcendental is empirical. Any object of awareness is empirical, so there is no awareness of transcendental freedom. Yet this argument is still strong enough to pose a threat to The Empirical Argument. In the introspective situation which way should a strict empiricist choose? When directly acquainted with an event which seems to him to be a cause but which does not seem to him to be an effect should he say it is a cause that is not an effect and that is why there is no experience of the cause of it, or should he say it is really an effect but only appears to be a cause that is not an effect because the cause of it is not revealed to experience? It is clear which way the strict empiricist should choose: he should provisionally adopt the view that there is at least one cause that is not an

Stephen Priest

Page 17

05/12/2007

effect but continue to look for causes of that cause by empirical means. Indeed, this is much the relationship between libertarianism in the philosophy of mind and neurology at present. But what the empiricist is thereby forced to concede is very close to transcendental freedom. He is holding as a provisional hypothesis that there is an event that is a cause but which is not itself caused, even though he would like to falsify that hypothesis empirically. It follows clearly that 'transcendental freedom' is not an empty concept in the sense of an uninformative one or one without a coherent epistemological function, even on a strictly empirical account of mental efficacy. Indeed, the empiricist needs something very like it to make sense of the situation he hopes will later be amenable to causal explanation. The criticism of concept empiricism in general would take us too far from the topic of free will so I do not pursue that here. But even without such criticism, the Empirical Argument leaves room for a weakened version of transcendental freedom and so does not amount to a disproof of 'I am free'. The Rules Argument The first premise is (a) 'Freedom is [[...]] a liberation from compulsion, but also from the guidance of all rules.' [Die Freiheit [] ist zwar eine Befreiung vom Zwange, aber auch vom Leitfaden aller Regeln] (A447, B475) The second is (b) 'If freedom were determined in accordance with laws, it would not be freedom.' [Wenn diese [d.i. Freiheit] nach Gesetzen bestimmt wre, sie nicht Freiheit [] wre.] (411, A447, B475) and the conclusion is; (c)'It is not permissible to say that the laws of freedom enter into the causality exhibited in the course of nature, and so take the place of naturallaws.' (A447, B475) [Denn man kann nicht sagen, da, anstatt der Gesetze der Natur, Gesetze der Freiheit in die Kausalitt des Weltlaufs eintreten.]

Stephen Priest

Page 18

05/12/2007

(Kemp Smiths reading of the German is not obviously right here. He might have reversed premise and conclusion. However, I shall go with Kemp Smith.) This argument is designed to rebutt any theory according to which freedom is governed by its own laws, which are putatively in some respects analogous to but in fact incompatible with laws of nature. If not plausibly synthetic a priori truths, the first two premises are either analytic, and so trivially true, or else synthetic and false. Certainly the first conjunct of the first premise is straightforwardly analytic. It is true by definition that freedom involves liberation from compulsion; this idea is essential to Kant's concept of transcendental freedom being uncaused causality. The second conjunct of the first premise is synthetic and false. It is clearly intended as synthetic because 'but also' is a grammatical marker which indicates that new and additional information is to be supplied not already expressed by the first conjunct. It is false because it is open to the same objection as the first two premises of the Argument From the Unity of Experience: from the fact that 'I am (transcendentally) free' is true it does not follow that my decisions and actions will not be rule governed. Of course these rules will not be the exhibiting of causal connections, nor rules according to which decisions and actions of a certain sort are shown to be consequences of causes of a certain sort. They may however be regularities or repetitions in decision or action. Here 'governed' or 'guided' cannot mean 'caused'. A non-Davidsonian libertarian might in addition claim that a person's reasons for performing an action were not causes of it; perhaps because they do not necessitate the action. Then it is open to such a libertarian to claim that one's actions exhibit a rule guided regularity that is not causal: it depends upon the contingent but uniform conjunction of roughly similar sorts of reasons being given for roughly similar sorts of actions. If this is right then 'I am free' need not require 'liberation from the guidance of all rules' The second premise is analytic if 'laws' means 'causal laws' because it would be self contradictory to assert that some event was transcendentally free, yet could be subsumed under a causal law. That would be the same as saying one and the same event both did and did not have a cause. But this analytic truth is one that the libertarian would be happy to concede. Construed synthetically however the claim is false because it is subject to the same sort of objection as the first premise's second conjunct. Freedom has to be distinguished from mere randomness or arbitrariness and it is possible that such regularities its exercise exhibits should not turn out to be causal. The conclusion, without the support of the premises, begs the questions against the libertarian. On Kant's criterion for the individuation of event-series an uncaused cause might 'enter into the causality exhibited in the course of nature' in just the sense of initiating a causal chain in which every subsequent event in the chain was a causal consequence of it, but where it was not itself a causal consequence of any event. This would not amount to the uncaused cause 'taking the place of natural laws', rather the chain thus initiated falls under natural law. On Kant's account, 'uncaused cause' has taken the place of 'caused cause'. If 'I am free' is true, and if this entails that there

Stephen Priest

Page 19

05/12/2007

are uncaused causes then human decisions are just such initiations of causal chains. (4) 2. Kants Compatibilism Kant's solution to the Third Antimony is transcendental idealism; the adoption of the distinction between things as they are in themselves, and as they appear to the human subject. Kant will argue that as noumenal selves we are free, but as we appear to ourselves we are determined in accordance with natural law. Here, if anywhere in the Critique of Pure Reason, Kant answers 'yes' to the question Am I free? The distinction between freedom and determinism which Kant draws (at CPR 464, A523, B560 - A533, B561) entails that freedom, like determinism, is a kind of causality: 'When we are dealing with what happens there are two kinds of causality conceivable by us: the causality is either according to nature or arises from freedom' [Man kann sich nun zweierlei Kausalitt in Ansehung dessen, was geschieht, denken, etweder nach der Natur, oder aus Freiheit] (CPR 464, A532, B560). Freedom is a kind of causality because a free event is a cause that is not an effect. Natural or determined causality in contrast requires that there be events which are both causes and effects. Kant uses this distinction to argue for the sort of freedom he thinks we do in fact possess. I Practical Freedom His best definition of 'practical freedom' is at CPR 464, A534, B562: 'Freedom in the practical sense is the will's independence of coercion through sensuous impulses' [Die Freiheit im praktischen Verstande ist die Unabhngigkeit der Willkr von der Ntigung durch Antriebe der Sinnlichkeit] I shall interpret this to mean: When presented with a choice between two courses of action X and Y, I may still do X instead of Y even if the closest possible empirical scrutiny of the causal conditions for my making the choice one way or the other suggested I would do Y. In other words, from no set of empirical sentences which form a prediction of my performance of a future action does it follows that I will perform that action. Further, it remains logically possible that I should perform an alternative action where the empirical evidence for such a performance ranges from

Stephen Priest

Page 20

05/12/2007

every small to nil. This formulation widens the scope of 'sensuous impulse' to (empirical) causal conditions'. This includes sensuous impulses in its denotation but, as we shall see, Kant needs' and often subscribes to, the wider formulation. My reading of 'practical freedom' allows Kant to retain his distinction between the animal will and the human will. Both human and (non human) animal wills are 'sensuous' because they are 'pathologically affected ie by sensuous motive' [pathologisch affiziert ist (durch Bewegungsursachen der Sinnlichkeit)] (CPR 465, A534, B562). This means that in the case of animals always but in the case of humans sometimes a particular action is causally determined by the mental state of the animal or human. All sensuous wills are sensuously 'affected' but only a subclass, the animal class have wills which are pathologically 'necessitated'. This is an important distinction between people and animals and I take it to be constitutive of people that they are selves, and being a self partially consists in having a will that is not entirely (pathologically) necessitated. The fact that Kant claims that the human will may be pathologically affected needs to be reconciled with its practical freedom. This can be done in a number of ways. It could be argued on Kant's behalf that if I am presented with two possible courses of action X and Y then I can always, in principle do X and not Y even if 'affected' to do Y. Even if I feel the 'sensuous impulse' to do Y I may still refrain from doing Y and do X . Alternatively, and less satisfactorily, I might sometimes be able to do X but sometimes affected to do Y rather than do X . This is less satisfactory because it suggests that persons may be pathologically necessitated, not just affected and that is something Kant denies. Further, it contravenes the principle that I may always do X even when a study of the empirical determinants of my behaviour suggest I would do Y. So, I think we have to take Kant literally when he says of the human will 'sensibility does not necessitate its action' [that Sinnlichkeit [] Handlungen nicht notwendig macht] (CPR 465, A534, B592). Here 'does not' means 'never does'. Although this is Kant's theory, it might be objected that such a strong doctrine of practical freedom is intuitively implausible. If the human will is never 'pathologically necessitated' then would it not be the case that I never lose when struggling with weakness of will? Such complete freedom from sensuous impulses would seem to make akrasia impossible. To meet this objection we have to allow with Kant that the human will is sometimes sensuously 'affected' where this implies that there are empirical causes of my action, and my felt real preference to do X rather than Y is not exercised. I lose the battle of the will, and my behaviour is empirically determined. Kant can allow this through his distinction between affected in general and necessitated in particular. I take it that if e necessitates e ' then it follows that e affects e ', but that if e affects e ' then it does not follow that e necessitates e '. This is because if e affects e ' then e (is amongst what) causes e ' to happen but it is further true in the case of necessitation that if e necessitates e ' then e ' could not not have

Stephen Priest

Page 21

05/12/2007

happened given that e happened. In contrast, if e affects e ' then it remains logically possible that e ' should not have happened even if e happened. The exercise of my freedom for example might have prevented the occurrence of e ' so long as e ' is affected but no necessitated, and that is precisely the way in which Kant has (here) drawn the distinction between human and animal wills. Although my actions may be causes they are not necessitated, but all the animal's actions are both caused and necessitated. It is unique to persons that they may perform some actions freely. After all, Kant talks of human freedom as a power and this does not commit him to the view that that power is always and invariably successful when exercised even though he retains the strong view that it could always in principle be so exercised. He says 'There is in man a power of self-determination, independently of any coercion through sensuous impulses.' [Dem Menschen [wohnt] ein Vermgen [bei], sich, unabhngig von der Ntigung durch sinnliche Antrieb, von selbst zu bestimmen.] (CPR 465, A534, B562) 'Self-determination' needs clarifying here. One way of reading it is this: e is self determining if e is the cause of itself. This is an obscure and barely coherent idea if it implies the logical possibility of something bringing itself into existence. In general if A causes B it follows that A is not identical with B, if only because causation is an n-ary relationship and there cannot be n-ary relationships without distinct relata. That would make 'A causes itself to exist' incoherent. But Kant does not need that possibly incoherent notion for his theory of freedom. We can read 'self-determination' in a way that allows persons to be (or be amongst) the causes of their own actions. Then 'self' determination refers to the person, so if I am self determining I cause myself to be what I am or I cause myself to do what I do. There is a problem about 'I' here, but in the last resort the referent of the first person singular pronoun for Kant will turn out to be one's free noumenal self. On yet another reading of 'self-determination' we may invoke the notion of uncaused causality outlined so far. Then if e is self-determining then e is a cause but not an effect: e is determining as a cause, but is self-determining in that it is not caused by anything other than itself to be thus determining. It is self-determining in the sense of 'autonomous' but not in the sense that it causes itself to be or to be what it is. The second two but not the first of these readings are close to Kant's use of 'selfdetermination'. The second in particular is essential to a conceptual link Kant thinks exists between transcendental and practical freedom. Transcendental and Practical Freedom

Stephen Priest

Page 22

05/12/2007

Transcendental freedom is an essential part of practical freedom, in the sense that if a person is practically free then that entails that that person is transcendentally free. The concept of transcendental freedom is 'the idea of a spontaneity which can begin to act of itself, without requiring to be determined to action by an antecedent cause in accordance with the law of causality' [die Idee von einer Spontaneitt, die von selbst anheben knne zu handeln, ohne da eine andere Ursache vorangeschickt werden drfe, sie wiederum nach dem Gesetze der Kausalverknpfung zur Handlung zu bestimmen] (CPR 465, A533, B561) Kant should have added here 'and in no way being determined to action by an antecedent cause' otherwise his formulation leaves it open that an event should be transcendentally free and yet causally determined which, as he admits, is selfcontradictory. This concept of transcendental freedom is deployed in two arguments: one designed to demonstrate the reality of practical freedom, that persons are practically free, and the other to show the impossibility of truly affirming the reality of practical freedom without thereby having truly affirmed the reality of transcendental freedom that persons are transcendentally free. In Kant's text these two issues are telescoped together so I shall try to separate them out. (a) The Argument for Practical Freedom

The premises are: (1) 'If all causality in the sensible world were mere nature, every event would be determined by another in time, in accordance with necessary laws' [Wenn alle Kausalitt in der Sinnlichkeit blo Natur wre, so wrde jede Begebenheit durch eine andere in der Zeit nach notwendigen Gesetzen bestimmt sein.] (2) 'Appearances in determining the will would have in the actions of the will their natural effects, and would render the actions necessary' [Da die Erscheinungen, so fern sie die Willkr bestimmen, jede Handlung als ihren natrlichen Erfolg notwendig machen mssten] (3) 'practical freedom presupposes that although something has not happened it ought to have happened'

Stephen Priest

Page 23

05/12/2007

[[Praktische Freiheit] setzt voraus, da, obgleich etwas nicht geschehen ist, es doch habe geschehen sollen] and the conclusion is (c) 'its cause (as found) in the (field of) appearance is not therefore so determining that it ex cludes a causality of our will.' [und seine Ursache in der Erscheinung also nicht so bestimmend war, da nicht in unserer Willkr eine Kausalitt liege. (CPR 465, A534, B562) In (c) 'its' refers back to 'something' in 3. (c) as it appears in Kant's text is a weaker conclusion than an affirmation of practical freedom. It is the claim that practical freedom is merely not excluded by natural law determinism. But the context of the argument, for example the straight affirmation at CPR 465, A534, B562 that 'there is in man a power of self-determination', makes it clear that Kant would like to draw the stronger conclusion, and perhaps he thinks he has done. To obtain the stronger conclusion we need to add some assumptions. Suppose we accept the truth of all three premises. Then we need to supplement (3) with this: (4) If something ought to happen it could even if it does not. The determinist cannot accept (4) but Kant clearly can. His distinction between affectation and necessitation gives him the distinction between what both happens and had to happen, and what happens but could, in principle, not have happened (given a particular set of empirical conditions). Then freedom is postulated as what makes this distinction possible. If this seems unacceptable to the determinist, three lines of objection are open to him. He can deny the distinction between what happens and what has to happen allowing only what could not not happen, but that is really to collapse natural law determinism into a kind of metaphysical fatalism. He could try to make sense of 'ought' without implying 'can' and that would need much extra argument. Or, finally, he could postulate something other than the freedom of the will that may operate 'contrary to the force and influence' of 'natural causes', which would hardly be more acceptable to the determinist than free will itself. Kant's argument is really quite a strong one. It amounts to: We need 'free' to make sense of 'ought', because 'ought to have' and 'ought not to have' imply 'could' and 'could have done otherwise'. So, if is true that we ought not to have done X then it is true that we could have not done X , or were free not to do X . Similarly, if it is true that we ought to have

Stephen Priest

Page 24

05/12/2007

done Y then it is true that we could have done Y or are free to do Y. The determinist either has to give up the language or morals or find a non-Kantian account of the truth conditions for 'ought' sentences. (b) The Argument from Practical to Transcendental Freedom (1) and (2) from the Argument for Practical freedom also feature as the first two premises here. The third premise is the definition of transcendental freedom: (3) 'Transcendental freedom is the idea of a spontaneity which can begin to act of itself, without (requiring) to be determined to action by an antecedent cause in accordance with the law of causality.' [[Transzendentale Freiheit ist] die Idee von einer Spontaneitt, die von selbst anheben knne zu handeln, ohne da eine andere Ursache vorangeschickt werden drfe, sie wiederum nach dem Gesetze der Kausalverknpfung zur Handlung zu bestimmen] (CPR 465, A533, B561) The fourth premise is the definition of practical freedom (4) 'Freedom in the practical sense is the will's independence of coercion through (sensuous) impulses.' [Freiheit im praktischen Verstande ist die Unabhngigkeit der Willkr von der Ntigung durch Antriebe der Sinnlichkeit] (CPR 465, B534, B562) and the conclusion is (c) 'The denial of transcendental freedom must, therefore, involve the elimination of all practical freedom.' [Die Aufhebung der transzendentalen Freiheit [wrde] zugleich alle praktische Freiheit vertilgen.] Is the argument sound? Soundness requires both validity and the truth of the premises, so let us assume the truth of deterministic premises (1) and (2) as before. Let us assume also the accuracy of Kant's definitions of 'freedom' in (2) and (4). Now the only question that remains is that of validity; whether (c) follows from the conjunction of (1)-(4). To decide this we need to know whether the truth of (1) and (2) precludes the truth of freedom in the sense of (4), and whether the nonexistence of freedom in the sense of (4) precludes, logically, the possibility of

Stephen Priest

Page 25

05/12/2007

freedom in the sense of (3). If these two conditions are met then (6) follows and the argument is valid. The existence of determinism as defined in (1) and (2) clearly is inconsistent with the existence of freedom as defined in (3) because (3) asserts that freedom is independence of causality, but (2) asserts that if determinism is true then the actions of the will are effects of causes. So, (3)'s freedom is incompatible with (1) and (2)'s determinism. If we give (4) the wider reading indicated above then denying (4) will entail denying (3) at least in this respect. (3) defines a sort of freedom from causality in general, (4) defines freedom of the will from causality in particular. So even if determinism is true there could be some sort of transcendental freedom, but this could not possibly be freedom of the will. (It could, for example, be the sort of transcendental freedom required to give sense to a certain account of the origin of the universe). But so far as the freedom of the will is concerned Kant's conclusion goes through: If I am not practically free, then I am not transcendentally free either. This in fact accords well with Kant's general thesis that all practical freedom requires transcendental freedom but not all transcendental freedom requires practical freedom. That is because not all freedom is freedom of the will (the putative uncaused first cause for example) but all practical freedom is freedom of the will. That is why 'will' appears where it does in the definition of 'practical freedom' (4) above. May Kant coherently state that persons are free as they are in themselves but are causally determined as they appear to themselves? He intends us to accept this thesis within the framework of transcendental idealism but before we can, several conceptual obstacles need to be removed. The first is this: 'Practical freedom presupposes [...] a causality which, independently of those natural causes, and even contrary to their force and influence, can produce something that is determined in the time order in accordance with empirical laws, and which can therefore begin a series of events entirely of itself.' [[Praktische Freiheit] setzt voraus [] eine Kausalitt [] [welche] unabhngig von jenen Naturursachen und selbst wider ihre Gewalt und Einflu etwas hervorzubringen, was in der Zeitordnung nach empirischen Gesetzen bestimmt ist, mithin eine Reihe von Begebenheiten ganz von selbst anzufangen.] (CPR 465, A534, B562) The problematic phrase here is 'contrary to their (natural causes) force and influence'. If it is true that everything phenomenal is causally connected and may be subsumed under natural laws, and if it is further true that causes necessitate their effects, how can it also be true that the kind of causality present in practical freedom may act 'contrary' to such causal connections and natural laws?

Stephen Priest

Page 26

05/12/2007

Kant has a solution to this problem but I think its force can only be fully appreciated if we accept the distinction made earlier between two sorts of description under which an event may fall, one featuring 'cause', the other featuring 'effect'. First it is useful to distinguish different possible strengths of 'contrary to' as it appears in 'contrary to their force and influence'. On the strongest interpretation, the exercise of freedom breaks a law of nature. Kant is not interested in such a strong sort of freedom. A weaker construal is this: If e and e ' are phenomenal events causally linked such that the occurrence of E is both a necessary and sufficient condition for e ' (ie such that unless e happened e ' could not happen, and such that if e happened then e ' could not not happen), then a free event f (ie a cause that is not an effect) operates in a way 'contrary' to e, and the connection between e and e ', if and only if the occurrence of e would prevent the occurrence [the occurrence] of e ' even if e happened. This is a strong interpretation of 'contrary' because it would seem that if e happens then e ' must happen, but if f happens e ' cannot happen, but it is further asserted that both e and f happen. What follows from this is logically impossible: that e ' both does and does not happen. Kant does not argue this through, but it is sufficiently threatening to his compatabilism to warrant a prima facie 'yes' in answer to his question: 'Is it a truly disjunctive proposition to say that every effect in the world must either arise from nature or from freedom?' [ob es ein richtigdisjunktiver Satz sei, da eine jede Wirkung in der Welt entweder aus Natur, oder aus Freiheit entspringen msse] (CPR 466, A536, B564) Here 'truly disjunctive' means we should read 'or' as aut and not vel, such that if there is freedom, there is not determinism and if there is determinism there is not freedom. Suppose we do not think of the exercise of freedom as preventing (per impossible) the occurrence of an event which is causally necessitated, but rather as causing it. Kant's solution is to adopt neither of these courses. Instead he makes use of the idea of practical freedom initiating a causal chain. The three strategies may be represented as follows: (i) e happens and f happens so e ' both does and does not happen (ii) e happens and f happens so e ' does happen (iii) F happens and some event other than e and e ' happens irrespective of whether e and e ' happen.

Stephen Priest

Page 27

05/12/2007

Strategy (3) is expressed in the following argument at CPR 467, A537, B565: (1) 'The effects [...] an intelligible causes appear, and accordingly can be determined through other appearances. [Obzwar ihre Wirkungen erscheinen, und so durch andere Erscheinungen bestimmt werden knnen]. (2) 'But its causality is not so determined' [Eine solche intelligible Ursache aber wird in Ansehung ihrer Kausalitt nicht durch Erscheinungen bestimmt] (3) 'While the effects are to be found in the series of empirical conditions the intelligible cause, together with its causality, is outside the series' [Sie ist also samt ihrer Kausalitt auer der Reihe, dagegen ihre Wirkungen in der Reihe der empirischen Bedingungen angetroffen werden.] (c) 'Thus the effect may be regarded as free in respect of its intelligible cause and at the same time in respect of appearances, as resulting from them according to the necessity of nature' [Die Wirkung kann also in Ansehung ihrer intelligblen Ursache als frei, und doch zugleich in Ansehung der Erscheinungen als Erfolg aus denselben nach der Notwendigkeit der Natur angesehen werden.] (c) expresses compatabilism. So, are (1), (2) and (3) true, and do they yield (c)? I interpret (1) as follows: If f is an intelligible cause then f is a free event, that is a cause that it is not an effect. It is also true of f that f can be known to exist by the exercise of the intellect (for example by following the argument above) but F is not itself a possible object of experience. For this I read 'intelligible' to mean 'only intelligible, and not also perceptible'. An effect of an intelligible cause, call it here 'ef'; is a possible object of experience. This gives a restricted reading to 'can be determined through other appearances' and 'appear(s)'. (2) just confirms the meaning of 'intelligible' as 'merely intelligible and not perceptible'. (3) is the claim that a free cause, a cause that is not an effect, is not embedded within an causal chain. Kant is in danger of making this claim rather too strong for the argument because what he needs is for f to be a cause of subsequent events which are indeed jointly a series of empirical events, but f must not itself be the effect of any chain or event in a chain of antecedent events. f is an unperceptible initiator of a chain of perceptible events.

Stephen Priest

Page 28

05/12/2007

Also, we should not read 'outside the series' to mean 'outside the series in time', only as outside the series in respect of being caused. This is because Kant argues that his, for example, rising from his chair, is chronologically a part of a series of empirical events, Even those preceding it are not causes of it if it is performed freely. There is no a priori objection to the truth of (1), (2) and (3) and he has already argued for (1). Also, (2) either qualifies, or is entailed by (1) depending on how we construe 'intelligible'. (3) is really a conjunction, the first conjunct of which (introduced by 'which') is logically entails (1), so long as the series of empirical conditions are appearances, which they are on Kant's account. But to derive (c) from the conjunction of (1), (2) and (3) it must be read in a special way: It must be possible to seem as though my actions are empirically caused by empirical conditions which are in fact events which merely chronologically predate my action and do not in fact cause it to happen in any degree. For the argument to go we have to read '(the effect may be regarded) in respect of appearances as resulting from them according to the necessity of nature' otherwise (3) falls into the incoherence of (i) above, or the vacuousness of (ii). Only that reading, whereby it is an illusion that my free actions are empirically determined, allows my actions to be uncaused events with empirical effects but without genuine empirical causes. This makes use of Kant's distinction between appearances and things in themselves in a special way. I show now that a different use will not work. Kant wonders whether 'An effect, not withstanding its being [[...]] determined in accordance with nature, may (not) at the same time be grounded in freedom.' [Es ist also nur die Frage: ob dem ungeachtet in Ansehung eben derselben Wirkung, die nach der Natur bestimmt ist, auch Freiheit stattfinden knne, oder diese durch jene unverletzliche Regel vllig ausgeschlossen sei] (CPR 466, A536, B564) For this to be the case some event Nevertheless, if e is necessary for e ' then f cannot be sufficient for e ' and if f is necessary for e ' then e may be sufficient for e '. It would have to be both necessary and sufficient for e ' and some free act F would have to be both necessary and sufficient for e '. But that seems hopeless. If e is sufficient for e ' then f is not necessary for e ', and if f is sufficient for e ' then e is not necessary for e '. (Nevertheless, if e is necessary for e ' then F may be sufficient for e ' and if F is necessary for e ' then e may be sufficient for e '. If there is overdetermination, it may be the case that e is both necessary and sufficient for e ' and that f is both necessary and sufficient for e '. It is logically possible if e exists if and only if f exists.) The solution I have just extracted from (c) obviates this difficulty. It asserts that f is both necessary and sufficient for e but that e only appears to be both necessary and sufficient to e ' but is in fact merely chronologically antecedent to e '.

Stephen Priest

Page 29

05/12/2007

That is the most sympathetic construal of his compatibilism: the one that allows him to present a solution to the problem of freedom and determinism. So it does allow as true 'I am free'. One other objection needs removing: If I am the free initiator of a causal chain, it follows that there takes place a series of empirical events which do not have an empirical cause. Is this intrusion into the phenomenal world logically consistent with the synthetic a priori claim that every empirical event has a cause, and does it violate any law of nature? As far as the first of these is concerned, an exception would have to be made of that empirical event which is the first effect of my free action if the synthetic a priori thesis is interpreted to mean every empirical event must have an empirical cause. But there seems no compelling need for Kant to make that stipulation. Even if some empirical events are caused by uncaused free actions, those empirical events are still caused and so do not violate the principle. Similarly with laws of nature, if a free action initiates a causal chain and that causal chain is empirical, ie, is a possible object of appearance and is embedded in a nexus of causal relations, no causal law has thereby been broken. Kant thinks natural laws hold only of phenomena, so if there is a noumenal and free cause that is not an effect no natural law is broken by that. Then there is nothing to preclude the event-components of the resultant causal chain effecting changes in the empirical reality in which it is embedded. This amounts to the fact that it is coherent to suppose that I may intervene in the course of nature, and allows us to give a final and very weak reading to 'contrary' above. This is: Some free action f is 'contrary' to some course of events in nature if those events would not have happened if f had not happened, or if those events would have happened unless f had happened. In this respect (though of course not as an effect) f has the same status as any other necessary and sufficient condition for an event or set of events. My Freedom I, as I am in myself, am a free agent. I as I empirically appear to myself and others am causally determined. But this raises a problem. If it is true that 'This acting subject would not in its intelligible character, stand under any conditions of time, time is only a condition of appearances and not of things in themselves.' [Dieses handelnde Subjekt wrde nun, nach seinem intelligibelen Charakter, unter keinen Zeitbedingungen stehen, denn die Zeit ist nur die Bedingung der Erscheinungen, nicht aber der Dinge an sich selbst.] (CPR 468, A539, B567) then it does indeed follow that

Stephen Priest

Page 30

05/12/2007

'In this subject no action would begin or cease.' [In ihm wrde keine Handlung entstehen, oder vergehen.] but then how can such a subject be coherently described as an agent at all? If I am capable of action then I must be capable of beginning to act and ceasing to act. My actions, in order to count as such, just last a certain time. One Kantian answer would be that the concept of an agent is an empirical one, and only in so far as I am a possible object of intuition may I be conceived as a (spatio-temporal) agent. This will hardly do though. The doctrine of transcendental freedom, implied by the doctrine of practical freedom, requires that I be the noumenal cause of effects in world as it appears empirically. The conjunction of the conclusion of the Transcendental Aesthetic, that the forms of intuition apply only to phenomena, with the claim that I am a noumenal cause of empirical events commits Kant to the view that there may exist what may be truly called a 'cause' yet to which no spatiotemporal predicates apply. I as I am in myself am such a cause. It might turn out that the conjunction of 'x possesses causal efficacy and x is not spatio-temporal' is not self-contradictory but first it is worth inspecting the motives which led Kant to this difficulty. If Kant admits that noumenal causes have spatio-temporal properties, then two gross inconsistencies with the remainder of his philosophy are introduced. Firstly the idea of a noumenon collapses into the idea of an empirical object, because if per impossibile noumena were spatio-temporal they would be possible objects of experience. Secondly, if noumenal causes were spatio-temporal they would be embedded in an empirical causal nex us and freedom in Kant's sense would be impossible. Kant is fully aware of these two threats. Indeed, the second is his motivation for emphasising the non-temporality of the numenal at CPR 468-9, A540, B568): 'In this subject no action would begin or cases, and it would not therefore have to conform to the law of determination of all that is alterable in time, namely that everything which happens must have its cause in the appearances which precede it.' [In ihm wrde keine Handlung entstehen, oder vergehen, mithin wrde es auch nicht dem Gesetze aller Zeitbestimmung, alles Vernderlichen, unterworfen sein: da alles, was geschieht, in den Erscheinungen (des vorigen Zustandes) seine Ursache antreffe.] He could in fact have drawn a stronger conclusion here than is implied by 'would not therefore have to', because if x is not temporal it is not clear how c could be empirically determined at all. The stronger conclusion is drawn in this passage:

Stephen Priest

Page 31

05/12/2007

'In as much as it is noumenon nothing happens in it; there can be no change requiring dynamical determination in time, and therefore no causal dependence upon appearances.' [So fern es Noumenon ist, nichts [in ihm] geschieht, keine Vernderung, welche dynamische Zeitbestimmung erheischt, mithin keine Verknpfung mit Erscheinungen als Ursachen angetroffen wird. ] (B569) Sensitive to the tension this introduces, Kant adopts two strategies by each of which he hopes there may be noumenal causes with empirical effects. The first of these relies on the thesis that nothing positive may be known about noumena anyway, and that if we do try to think them our thinking if confined to empirical categories: 'This intelligible character can never, indeed, be immediately known [...] it would have to be thought in accordance with the empirical character just as we are constrained to think a transcendental object as underlying appearances, though we know nothing of what it is in itself.' [Dieser intelligibele Charakter knnte zwar niemals unmittelbar gekannt werden [] aber er wrde doch dem empirischen Charakter gem gedacht werden mssen, so wie wir berhaupt einen transzendentalen Gegenstand den Erscheinungen in Gedanken zum Grunde legen mssen, ob wir zwar von ihm, was er an sich selbst sei, nichts wissen.] (CPR 468, A540, B568) Yet elsewhere Kant insists that it is known as a fact about noumenal selves that they are practically, and hence transcendentally free. (See for example: CPR 476, A554, B582). Clearly if I am transcendentally free and so the uncaused cause of empirical effects, it must be true of me that I am a noumenal cause. This cannot just be a way of thinking of ourselves that we are forced to adopt to make sense of our moral language. If I am free in Kant's sense it must be true that I am a noumenal cause. Kant realises that the 'no positive knowledge of noumena' strategy fails, so adopts this one; 'No action begins in this active being itself; but we may yet correctly say that the active being of itself begins its effects in the sensible world.' [Man wrde von ihm ganz richtig sagen, da es seine Wirkungen in der Sinnenwelt von selbst anfange, ohne da die Handlung in ihm selsbt anfngt.] (CPR 469, A541, B569)

Stephen Priest

Page 32

05/12/2007

I take it this means that the noumenal self is a cause of a variety of empirical effects without itself either beginning ending or changing. Is this supposition coherent? I leave aside the conceptual problem about whether c may exist without x lasting at all. Let us assume that noumenal selves may exist whether or not they have duration. Then, may a noumenal self initiate spatio-temporal changes without changing itself? This seems to me not logically impossible, that is, its description does not entail a contradiction, but it is rather hard to imagine. We have to imagine that a noumenal self, call it x, does not come to exist or cease to exist and does not change, yet causes a series of numerically and qualitatively distinct events to occur in the empirical world. The imaginative difficulties are these. If x does not change, there seems to be no possible fact about x which would make x initiate e [...] en, rather than, say, e[...]..e'n. Also, if x is not temporal may x initiate any temporal event series at all? Suppose x lasts, then that would allow x to last longer than t-t3, because x does not begin or end. If x lasts, in other words, x is eternal (ie, is, has always been and will always be). It need not be a difficulty for Kant's theory that there are no facts about C that would ex plain why x caused e[...] e'n, or say, ee[...] en, rather than e[...] en. This is not just because if we are ignorant of noumenal facts we will be ignorant a fortiori of facts about the causal nature of noumena, but also because the thesis accords well with Kant's notion of free action being a kind of spontaneity. If we could per impossibile discover noumenal facts which explained why one empirical chain was initiated, or initiated rather than another, then there is a sense in which we could no longer be talking about free actions at all. Transcendental freedom is a very strong sort of freedom, whereby there are no causes of my free actions. If there could be a noumenal discovery of why I initiated a certain causal chain, then there would be a severe danger than such an explanation would be a causal one. This need not be a threat to a weaker libertarianism than Kant's, say where my freedom consists in my decisions and motives being (among) the causes of my actions, but Kant could not possible accept this, just because there are no causes of transcendentally free actions. The upshot of this is that there is no a priori objection to my noumenal self being an uncaused cause of a variety of empirical events even if such random spontaneity seems too much like chance to accord much with our ordinary intuitions about our freedom of action. There is a different sort of problem about noumenal causality. If noumenal causes are not possible objects of experience, and if they can be known only intellectually, and if it is further true that the empirical effects of noumenal causes are temporally preceded by earlier empirical events (which are not their causes) what difference could there be between an empirical event which spontaneously initiated an empirical causal chain, and am empirical event that was noumenally caused to do the same? The straightforward answer is just that in the first case there is no noumenal causation and in there second there is. But what are the grounds for such a

Stephen Priest

Page 33

05/12/2007

distinction in our thinking? Kant realises the danger: When the noumenal self 'of itself' initiates a causal chain 'we should not be asserting that the effects in the sensible world can begin of themselves.' [und dieses wrde gltig sein, ohne da die Wirkungen in der Sinnenwelt darum von selbst anfangen drfen] (CPR 469, 541, B569) Clearly it is a logical impossibility that any kind of effect may begin of itself, but not that any event should begin of itself. On a strongly anti-realist construal there would not appear to be any difference between an unobservable and merely thinkable cause of an empirical event, and no cause at all. Instead of facing this difficulty square on Kant makes this slide: 'they ('the effects in the empirical world') are always predetermined through antecedent empirical conditions' [weil sie in derselben Zeit, aber doch nur vermittelst des empirischen Charakters [] vorher bestimmt, und nur als eine Fortsetzun der Reihe der Naturursachen mglich sind.] If 'determined' means 'caused' here then that is false, because if an empirical event is the first event in a causal chain initiated by a noumenal cause then that event cannot have any empirical cause. But if 'determined' means either of 'appears to be caused' or 'is temporally antecedent to' then the statement accords with the best reading of Kant's theory. On either account however, this does not meet the anti-realist objection. The way to meet it on Kantian grounds is to postulate freedom transcendentally as the conclusion of the 'ought' argument above. Then the difference between 'we are transcendentally compelled to think of ourselves as free' and 'we are free' collapses. In other words, there is no room to argue for a sceptical difference between how the world is, and how we must conceive it. Allison, Guyer, Irwin In Kant and the Claims of Knowledge Paul Guyer levels a series of criticisms at Henry Allisons Kants Theory of Freedom (Allison, 1990). I offer a way of reconciling the positions of Allison and Guyer. Guyer quotes Allison: to conceive of oneself (or someone else) as a rational agent is to adopt a model of delberative rationality in terms of which choice involves both a taking as and a

Stephen Priest

Page 34

05/12/2007

framing or positing. Since these activities, as expressions of spontaneity, are themselves merely intelligible (they can be thought but not experienced) it is necessary to attribute an intelligible character to the acting subject. (Allison, 1990: 38) and Guyer reports Allisons position: Because of their spontaneity, any acts seen as products of practical freedom (action according to a principle) must also be seen as products of transcendental freedom (action free of determinism) (Guyer, 1992: 103) Guyer accepts from Allison that Kant may well suggest such a position in the first Critique (see A 546-8 / B 574-6). but Guyer criticises Allison for endorsing Kants argument, for the following reasons: [1] Even so, it is clearly a poor argument. Conceptualisation and judgment may well be spontaneous rather than receptive in that they are no products of sensory stimulation alone but emerge only from internal cognitive processing of sensory stimuli; but that does not imply that there is anything nonnatural or nondeterministic about such cognitive processing. (Guyer, 1992: 103) [2] although incorporating a desire into ones maxims is active in a way that the mere occurrence of the desire is not, it hardly follows that this activity itself cannot be ex plained within a naturalistic framework. (Guyer, 1992: 103) [3] Without a great deal more argument, the contrast between spontaneity and passivity is only an unredeemed metaphor that cannot support an inference to the real existence or even the mere possibility of a breach with determinism. But this contrast is all that Allisons defense of the spontaneity thesis amounts to; he does not attempt to prove that mental activity is really incompatible with determinism. (Guyer, 1992: 103) Guyer attacks Allison for adopting a two-aspects interpretation on Kant on phenomena and things-in-themselves: [1] No disturbing metaphysics but only the comfortable conceptual distinction between two points of view is therefore necessary to justify Kants assertion of the compatibility of empirical determinism and transcendental freedom. (Guyer, 1992: 104)

Stephen Priest

Page 35

05/12/2007

[2] Kant doubtless suggests such an interpretation of his transcendental idealism at various points in the first Critique (e.g. B 307), but there are also places where his arguments actually imply the stronger view, not merely that our concepts of things in themselves exclude space and time but that the things themselves are not spatial and temporal (e.g. A 26 / B42). (Guyer, 1992: 104) [3] It would seem to be the latter rather than the former interpretation that is necessary to make sense of his claims about noumenal freedom. (Guyer, 1992: 104) If the phenomenal self is not the noumenal self and if the phenomenal self is determined but the noumenal self is free, nothing is thereby both determined and free. The inconsistency between freedom and determinism is therefore avoided, or remains only formal. The price of this reconciliation is high: I am two numerically distinct selves. Nevertheless, Guyer might not be right that endorsement of the twoworlds ontology is necessary for a reconciliation of transcendental freedom with determinism. I agree with Guyer that the situation seems hopeless. However, consider this strategy: [A] Assume the double-aspect ontology (following Allison, Bird (1962) et. al.) and reject the two-worlds ontology. [B] Assume that if p appears to be the case p might be true or p might be false. [C] Consider what is qua appearance: If an agent or action is considered: (1) It appears determined. (2) It is not determined. (3) It is really free (but not qua appearance). If anything that is not an agent or action is considered: (1) It appears determined. (2) It is determined. (3) It is not really free (at any level). [D] Consider what is as thing-in-itself: If an action or agent is considered: (1) It does not appear but must be thought of as free. (2) It is free. (3) It is not determined (on any level). If anything that is not an agent or action is considered:

Stephen Priest

Page 36

05/12/2007

(1) It does not appear. (2) It is not free. (3) It is not determined. (4) It is only minimally thinkable (eg. that it is is thinkable). The interpretations amassed under [C] and [D] are mutually consistent and so provide a coherent reading of Kants compatibilism which does not ascribe to him the two-worlds ontology. As with most compatibilisms, something large has to be given up. In this case, it is the view that determinism is any more than apparently true of human agents. The price of this solution is that Kant is not really a determinist and so not really a compatibilist. On the other hand, his theory comes out as consistent and to that extent, possibly true. If we take noumenal freedom seriously the reading also makes the theory sound quite Kantian. It is a reading of the Critique of Pure Reason which grounds the concept of freedom in the moral philosophy. In this dispure, Terence Irwin agrees with Guyer: Now if an event is determined, it is true of it under all descriptions that it is determined, even though only some true descriptions [] show why it is determined (Irwin, 1984 quoted in Allison, 1990: 38, quoted in Guyer, 1992: 105) and Guyer defends Irwin against Allisons reply. Here I adjudicate between, on the one hand, Guyer and Irwin and, on the other hand, Allision. [1] Allison is wrong to argue that Irwins attack rests on the assumption that being causally determined is a necessary or essential property of any occurrence of which it is predicated (Allison, 1990: 44) and to argue that the concept of causality is merely an epistemic condition, more specifically, a condition of the representation of temporal change. (Allison, 1990: 44) [2] Guyer is right to argue that: Irwins point is simply that, if determinism is in fact true at all of humans as they actually are, then it remains true of them no matter how they are described, for leaving determinism out of some description does not make it false. (Guyer, 1992: 105) Irwin has pointed to a problem which appears fatal to the two-aspect solution to the problem of freedom. However, this problem can be circumvented. It requires giving up the assumption that agents or actions are determined on the Kantian view. Of

Stephen Priest

Page 37

05/12/2007

course this invites the objection: They are determined. They are phenomenally determined, therefore they are determined. But what does Phenomenally determined mean here? It means appear to be determined and that is consistent with not determined. This solution involves giving up the assumption that the inference from: (1) a is phenomenally F to (2) a is F is valid. As a matter of Kantian exegesis this inference should be given up. One point of Kants reiteration of the claim that there are appearances not just things in themselves is that appearances might not be veridical. We should not conversely assume they are always and everywhere non-veridical. The inference from (1) a is phenomenally F to (2) a is not F is also invalid. Guyer says: Allison is forced to maintain that determinism must be true of our representations of ourselves but false or at least possible false of ourselves as we really are. (Guyer, 1992: 105) But we can give up the assumption that determinism is true of us at any level. Then the two-aspect solution is viable. Guyer also says: So Allison cannot maintain his interpretation of transcendental idealism yet answer Irwins objection (Guyer, 1992: 105) The solution is to give up an assumption that alllows the Irwin objection to be formulated. Give up: We are determined. (5) Freedom and Self-Knowledge The distinction between phenomena and noumena is also relevant to a view Kant holds about the epistemology of freedom; about how it is possible for one to know of

Stephen Priest

Page 38

05/12/2007

oneself and others that we are free. He does not take the clearly implausible view that people are acquainted in general with philosophical arguments for freedom on which they base their belief, yet, Kant rightly notes that most people are intuitive libertarians. Kant has two accounts of how it is possible for us to know we are free. One to do with pure apperception, the other to do with rationality. (1) Freedom and Pure Apperception There is a sense in which Kant thinks first person singular knowledge of freedom is epistemologically prior to third person knowledge of freedom. This is because he thinks there is an asymmetry between one way of obtaining knowledge that a person may possess with regard to himself and others, and another sort of access to knowledge that a person may possess only with regard to himself. This is not at all to say that there are two sorts of freedom. 'Freedom' means the same in 'I am free' as in 'He, she is free'. It is just to say that on one of Kant's accounts knowledge, sentences of the second form is only possible on condition there is knowledge of the first form. I thus can only know He is free is true if I know myself to be free. (This is part of what has come to be called Kant's 'Cartesian starting point'). Here is one means of acquiring knowledge about oneself and others. We could call it the observational route, and freedom cannot be known to exist only by its exercise: 'If we could exhaustively investigate all the appearances of mens wills there would not be found a single human action which we could not predict with certainty, and recognise as proceeding necessarily from its antecedent conditions. So far, then as regards this empirical character there is no freedom.' [Wenn wir alle Erscheinungen [des Menschens] Willkr bis auf den Grund erforschen knnten, so wrde es keine einzige menschliche Handlung geben, die wir nicht mit Gewissheit vorhersagen und aus ihren vorhergehenden Bedingungen als notwendig erkennen knnten. In Ansehung dieses empirischen Charakters gibt es also keine Freiheit.] (474, A550, B578) Here is the first person singular privileged access to the fact of one's own freedom through pure apperception; 'Man, however, who knows all the rest of nature solely through the senses, knows himself also through pure apperception; and this indeed, in acts and inner determinations which he cannot regard as impressions of the senses. He is thus to himself on the one hand phenomenon, and on the other hand, in respect of certain faculties the action of which cannot be ascribed to the receptivity of sensibility, a purely intelligible object.'

Stephen Priest

Page 39

05/12/2007

[Allein der Mensch, der die ganze Natur sonst lediglich nur durch die Sinne kennt, erkennt sich selbst auch durch Apperzeption, und zwar in Handlungen und inneren Bestimmungen, die er gar nicht zum Endrucke der Sinne zhlen kann, und ist sich selbst freilich eines Teils Phnomen, anderen Teils aber, nmlich in Ansehung gewisser Vermgen, ein blo intelligibeler Gegenstand.] (CPR 472, A546, B574) Despite his empirical realism, it is therefore not at all Kant's view that the observational route can yield everything that can possibly be known about a person, yet the various scientific methods of studying human beings are all constrained by it: 'It is only in the light of this (empirical) character that man can be studied - if that is to say we simply observing, and in the number of anthropology seeking to institute a physiological investigation into the motive causes of his actions.' [Nach diesem knnen wir doch allein den Menschen betrachten, wenn wir lediglich beobachten, und, wie es in der Anthropologie geschieht, von seinen Handlungen die bewegenden Ursachen physiologisch erforschen wollen.] (CPR 474, A550, B578) First I make some remarks on Kant's statement of the observational route. It takes the form of an argument: If (1) we had complete empirical knowledge of a person then (2) we could predict that person's actions with certainty. Therefore (c): That person is not free. The argument, although frequently deployed by determinists, is invalid. The faulty inference is from (2) to (c). (Let us accept the truth of (1) and (2) and the validity of the inference from (1) to (2).) The reason why it does not follow that a person is not free from the fact that that a person's actions may be predicted in detail is that the basis of the prediction might in fact be a regularity with which that person exercises their freedom of choice. It then begs the question if the determinist insists that he has discovered all the empirical determinations of that persons behaviour.

Stephen Priest

Page 40

05/12/2007

'Appearances of mens wills' is vague in Kant's formulation but there is nothing about it which precludes those appearances being the regular effects of free actions. Indeed, this is really the conclusion Kant wants because he thinks there is both empirical knowledge of the appearance of determinism, and knowledge of freedom through pure apperception. If I know I am free, then, this cannot be on the basis of my observing myself or others. In this sense 'freedom' is not an empirical concept (even though Kant allows it has uses in empirical sentences). Rather through pure apperception, which is the capacity I have for rational non-empirical thought about myself - I can know that there exist 'inner determinations' which I cannot regard as sense impressions, and certain 'faculties of action which do not belong to sensibility. Kant is claiming that I have direct intellectual knowledge of the operations of my will. This is completely distinct from the findings of inner sense introspection. Pure apperception is rational or intellectual and contains no experiential dimension. Is there any evidence that such a capacity for knowing freedom exists? I do not see that there is. In Kant's text there is no basis for a distinction between knowledge of the reality of freedom through the exercise of the faculty of reason in general and pure apperception in particular. I do think of myself as free but this is because of the conclusion of an argument about moral language and the possibility of imperatives that extends to all persons, not just myself. I conclude that I in particular am free as an inference from the generalisation that persons are free. I do not acquire this knowledge on the basis of a private intellectual faculty that reveals to me only my own freedom. There is then no evidence provided by Kant for the claim that I do have or exercise any such faculty. Pure apperception has not been given any convincing role in acquiring knowledge of freedom independently of reason in general. Bennett on the Third Antinomy I offer a reading of Kant on freedom which differs in several important respects from that offered by Jonathan Bennett Chapter 10 of his Kant's Dialectic. I point out some contrasts and present some grounds for preferring my reading. Bennett thinks 'both sides of the third antinomy assume that whatever happens is caused to happen' (KD 184), but this leaves no room for Kant's concept of an absolute spontaneity, the idea that there logically might be an event which is a cause but not an effect, and without crediting Kant with this concept in the thesis and proof of the third antinomy little sense can be made of the doctrine of transcendental freedom. Indeed, it is because Bennett does not detect this notion in Kant's text that he thinks the thesis argument 'offers no positive doctrine of freedom' (KD 184), and ends on page 187: 'I confess defeat; I cannot discover how the thesis argument of the third antimony is supposed to work'.

Stephen Priest

Page 41

05/12/2007

What has gone wrong? Bennett notes correctly that Kant is discussing two putatively distinct forms of causality, one that operates in accordance with natural law, and one that allegedly operates freely, and is thus not constrained by natural law. This is what is right about Bennetts reading: that freedom is a kind of causality. But it cannot be inferred from that fact, even when conjoined with the fact that there is causality in accordance with natural law that there is no event that was not caused to happen. The slide is made from 'there are different sorts of causality, among them freedom' to 'there are no uncaused events', yet, clearly, complete determinism cannot be logically derived from an enumeration of the sorts of causality that might exist. There are two passages where I think Bennett comes close to extracting a Kantian doctrine of freedom as 'spontaneity' from the thesis. He seizes of Kant's 'cause (which is) sufficiently determined a priori' (KD 185), but then confesses 'I do not know what that means, but it cannot mean the same a sufficient cause' (KD 185), and reason he gives is 'for the latter points rather to the idea of a effect which is sufficiently determined' that Bennett needs in making sense of Kant's concept of freedom is a distinction between two semantically distinct description under which one and the same event might fall. Then we can say an event is both cause and effect: the cause of certain events which are chronologically subsequent to it, and the effect of certain events which are chronologically antecedent to it. Then it will still follow that 'caused cause' and 'caused effect' will not mean the same, but if we accept a distinction between sense and reference this need not matter much. There is at least then no incoherence introduced into the idea of one and the same event being a necessary and sufficient condition for the occurrence of a second event and some earlier event's occurrence being a necessary and sufficient condition for the occurrence of that first event. This formulation (which I think is close to Kant's) leaves open two logical possibilities, one of which is essential to Kant's formulation of the thesis argument, but which is almost wholly neglected by Bennett, the other of which is not considered by either Kant or Bennett. The first possibility is: (1) There might be causes which are not effects.

and the second is: (2) There might be effects which are not causes.

We may leave aside the second as of little consequence of the free-will issue (even though it might be a useful concept for making sense of the end of the universe). The first though must be true if 'I am free' is true, on the Kantian interpretation of 'freedom'. Bennett perhaps allocates too much scope to 'nothing' in Kant's '[[...]] nothing takes place without a cause sufficiently determined a priori'. (KD 185, CPR 410, A446, B474) 'Nothing' here covers 'no event which is in principle a possible object of appearance' and not 'nothing whatsoever'. This interpretation is born out of

Stephen Priest

Page 42

05/12/2007

Kant's immediately prefacing it with; 'the law of nature is just this'. We know from the Transcendental Deduction and the Principles chapter that laws of nature hold for the phenomenal world and not for things in themselves, or things as they are in themselves. This leaves room for events considered at the level of things in themselves not to be effected, even if it is an all pervasive feature of the phenomenal world of our experience that causal relations invariably obtain between the events occurring in it. It also leaves the logical possibility which Kant theory of freedom requires: that a noumenal cause may have a phenomenal world, but I might not thereby myself be the effect of any cause. If we read Kant this way we need not find his use of a priori puzzling here. Bennett says: 'The phrase 'a priori' seems not to carry the Kantian technical sense of 'independently of all experience, but rather the pre-Kantian sense of 'in advance' or 'independently.' (KD 185) If we take seriously the claim that 'cause' is a category, then the sentence 'every event has a cause' is a priori in just the sense of a priori that Bennett precludes here but which Kant intends. It is an a priori fact about appearances: about what our experience must necessarily be like, given the categorical framework we do posses. But clearly this is in no way inconsistent with Kant's theory of transcendental freedom because it is an important part of 'transcendental' to mean 'not a possible object of experience' or 'not an item amongst appearances'. So, if we accept a distinction between phenomena and things in themselves, and accept a distinction between the sense of 'cause' and the sense of 'effect' but assume a common referent for the two concepts, then Kant's doctrine of transcendental freedom may be extracted from the third antinomy in a way that is not open on Bennett's reading. The other passage where Bennett comes close to such a reconstruction is on page 184 (KD) when he says the thesis contains mention of 'an alleged causality in which the cause does not antedate the effect'. This can be read in conjunction with Bennetts claim a little later that 'there is another kind of causality - that of freedom in which an event's cause does not precede the event itself'. (KD 187) The first formulation leaves it open as a logical possibility that there could be an effect which predated its cause. Whether this is a possibility need not concern us here as it does not concern Kant. It can also be read to mean much the same as the second claim however. Then it means; there could be a cause that was simultaneous with its effect. The trouble with reading this notion into the third antinomy is that it forces Kant to make an assumption which is quite absent from the text. Kant is not at all dealing with the notion of two numerically distinct but simultaneous events one of which is the cause of the other, yet Bennett makes this constitutive of the kind of causality that is freedom. Now, it is difficult to see why anyone should make this part of the concept of freedom, and in fact Kant does not. From the fact that there exist two numerically distinct events e and e', where e is the cause of e' and yet

Stephen Priest

Page 43

05/12/2007

where both e and e' endure only from some earlier time t to some later time t', it does not follow that something constitutive of freedom in Kant's (or anyones) sense has thus been described. On the contrary, it might be consistent with the strongest sort of determinism that some cause and some effect may be simultaneous. Why then does this in any way come close to Kant's intention? Suppose an event is both a cause and an effect, in the admissible and Kantian sense advocated so far. Suppose further that two putatively numerically distinct events, a cause c and an effect ef, are in fact one and the same event 'e', then it will follow from this that c and ef last just as long as each other. Indeed, if it is a necessary condition for two putatively numerically distinct events really being one and the same event that they exist over all and only the same time, then it will be necessarily true that c and ef exist simultaneously if eand ef are in fact (numerically) identical. Bennett's account captures the possible simultaneity of cause and effect in Kant's theory, but not their possible identity. Still, I should not wish to make this identification of cause and effect constitutive of Kant's concept of freedom. On the contrary, that concept requires there to be some causes that are not identical with any effects. Noumenally, I am one of those. A further strong contrast between Bennett and myself is this. Bennett thinks the discussion of the possibility of a first event in the thesis is entirely irrelevant to the question of human freedom: 'If the concept of freedom solves problems about humans, then why invoke cosmology in its defence? If it cannot satisfactorily solve such problems, how could cosmology rescue it?' (KD 189) There are two connections that Kant would wish to insist upon that Bennett neglects. The first is, by demonstrating the possibility of a first event that is a cause Kant shows the logical possibility of an uncaused cause. This gives him as coherent just the concept he needs to demonstrate the possibility of human freedom. Admittedly, Kant need not have pursued this strategy. It does not follow that if there was an uncaused cause which was the first event, then there are subsequent uncaused events which are the free acts of humans in initiating causal chains. But Kant does not claim the obtaining of that strong condition anyway. Kant's procedure in the third antinomy is rather the one outlined by Strawson: 'Once causality through freedom is admitted, it is permissible to entertain the thought that other causal series, occurring within the course of the world's history may be similarly initiated through freedom' (BS 208) Or, as I should say, once the logical possibility of one uncaused cause has been demonstrated, then the possibility of uncaused causes or the coherence of 'uncaused cause' has been demonstrated. If we wish, we may take the discussion of the origin of the universe as a thought-experiment to make more palatable to us the logic of Kant's discussion of the freedom of the self.

Stephen Priest

Page 44

05/12/2007

The second connection relies on accepting transcendental idealism. Kant's phrase 'condition of appearances' is ambiguous between empirical realist and a transcendental idealist reading. Empirically a first cause is a condition of appearances if those appearances could not exist unless ultimately caused to exist by a first cause. But, transcendentally speaking, series of appearances could not exist (at least not qua appearances) unless they were presented to a perceiving subject, and for Kant this means a noumenally free rational subject. Bennett has some interesting remarks on the solution to the Third Antinomy. Bennett thinks Kant's compatibilism reconciles freedom and determinism at the cost of making freedom vacuous. This is because if concepts 'are tools for the orderly management of intuitions, then the theory of noumenal freedom does not even make sense' (KD 194) and 'That is the reconciling endeavour in a nutshell: the two sorts of causality cannot conflict because they cannot meet. One is confined to the empirical realm, whereas the other is banned from it' (KD 194) Admittedly, if 'cause' is a category, and categories have only an empirical use, then it makes no sense to speak of noumenal or purely rational causality. But Kant is not so much of an empiricist as that. It is a pervasive feature of the Critique of Pure Reason that much of the language which ostensibly only has the legitimate role of making our experience intelligible to us is in fact given the extra role of exhibiting the transcendental conditions for that experience. Kant thinks freedom is amongst those conditions: In particular we need it to make sense of our moral experience. But it is the second quotation which points up the difference between Bennett's interpretation and mine. On my reading it is just false that the two sorts of causality cannot meet. This is because Kant thinks there are noumenal causes with empirical effects: chains of empirical events which would not have happened without being caused by noumenal, free and rational acts of will. Whether or not this theory is true, or in the last resort fully intelligible is a separate matter from whether Kant subscribed to it, and the textual evidence is that he did. Bennett says about (A 572/B 580): Kant implies that intelligible causality produces or mediates natural causality; but this of which I can make nothing seems not to be his considered view. (Bennett, 1974: 193 fn 15) Intelligible causality is noumenal causality. Natural causality is determinism in the empirical world. I suggest that Kant means: the obtaining of intelligible causality is a necessary condition for the obtaining of natural causality. In particular, the way

Stephen Priest

Page 45

05/12/2007

anything is in itself is necessary for its empirical appearance. I am not suggesting that Kants intermittent endorsements of noumenal causality are consistent with his inclusion of causation in the list of categories, any of which has only an empirical use. Therefore this perhaps should not be his considered view. Nevertheless, if Kant allows a glimpse of the noumenal anywhere, it is in his theory of freedom: Noumenal ontology makes the phenomenal world possible. The epistemological order of priorities is the reverse: The intelligibility of noumenal freedom depends upon the applicability of the categories to phenomena. Bennett calls up in support of his interpretation Kant's statement that 'The inevitable consequence of obstinately insisting upon the reality of the empirical world is to destroy all freedom' (CPR 565) Again, Kant is not as empiricist as Bennett would like. We may, if we wish, read this with Bennett as asserting the existence of a neo-rationalist would of noumena which are more real than empirical appearances. If we adopt a more restricted reading, however, as a above, then Kant can be taken as warning us not to assume too dogmatically that human actions are the inevitable outcomes of deterministic chains of empirical events which predate them. It might be that they just postdate them. (6) 3. Kant on the Freedom of Reason Kant thinks that 'our reason has causality, or [[...]] we at least represent it to ourselves as having causality' [Da diese Vernunft nun Kausalitt haben, wenigstens wir un seine dergleichen an ihr vorstellen, ist [] klar] (CP.R 472, A547, B575) He has an argument for this, but before ex amining it we should decide what it means to say reason has causal efficacy. Kant is talking about practical reason, not about theoretical reason as this statement a few pages later makes clear: When we consider [[...]] actions in relation to reason - I do not mean speculative reason, by which we endeavour to ex plain their coming into being, but reason in so far as it is itself the cause in producing them.' [Wenn wir aber eben dieselben Handlungen in Beziehung auf die Vernunft erwgen, und zwar nicht die speculative, um jenen Ursprunge zu erklren, sondern ganz allein, so fern Vernunft die Ursache ist, sie selbst zu erzeugen.]

Stephen Priest

Page 46

05/12/2007

(CPR 474, A550, B578) We need to locate this notion of reason within the theory of practical freedom as interpreted so far. On that theory, if I am practically free then I am transcendentally free, and if I am transcendentally free then I am a causes but no thereby an effect of empirical events. If we now ask; 'what is that cause?' we can answer 'reason in its practical sense'. It would be tempting to straightforwardly substitute 'reasons' here and develop further the quasi Davidsonian identification of reasons with causes, but that would be tangential to Kant's present concern. Rather, I freely cause empirical events to happen because I believe they ought to happen, and it is the existence of that imperative which is identical with the cause which is my free action. It is in that sense that reason itself has causal efficacy. If on Davidsonian lines we wished to say that is my reason for initiating a certain empirical chain there is nothing on Kant's theory to preclude our doing so. His argument for the causal efficacy of reason is this: (1) 'Ought' expresses a kind of necessity and of connection with ground which is found nowhere else in the whole of nature.' [Das Sollen drckt eine Art von Notwendigkeit und Verknpfung mit Grnden aus, die in der ganzen NAtur sonst nicht vorkommt] (CPR 472-3, A547, B575) (2) 'Natural conditions [[...]] do not play any part in determining the will itself.' [Diese Naturbestimmungen betreffen nicht die Bestimmung der Willkr selbst] (CPR 473, A548, B576) (3) 'This ought expresses a possible action the ground of which cannot be anything but a mere concept.' [Dieses Sollen nun drckt eine mgliche Handlung aus, davon der Grund nichts anders, als ein bloer Begriff ist] (CPR 473, A547, B575) (4) 'actions (may be) necessary even though they have never taken place, and perhaps never will take place.' [und nach denen sie so gar Handlungen fr notwendig erklrt, die doch nicht geschehen sind und vielleicht nict geschehen werden.] (CPR 473, A548, B576)

Stephen Priest

Page 47

05/12/2007

(c) 'It (is) at least possible for reasons to have causality with respect to appearances.' [Da die Vernunft in Beziehung auf sie Kausalitt haben knne] (c'') 'That our reason has causality, or that we at least represent it to ourselves as having causality is evident form the imperative which in all matters of conduct we impose as rules upon our active powers.' [Da diese Vernunft nun Kausalitt habe, wenigstens wir uns dergleichen an ihr vorstellen, ist aus den Imperativen klar, welche wir in allem Praktischen den ausbenden Krften als Regeln aufgeben] (CPR 472, A547, B575) Premise (1) is a version of Hume's claim that from no set of non-evaluative empirical sentences may a value judgement be logically derived or, as it is often put, that it is not possible to derive an 'ought' from an 'is'. The kind of necessity Kant claims in (1) is not either logical or natural necessity. This is made clear by (4), because if P is logically necessary then P could not fail to be true, but (4) suggests there may be necessities in a non-logical sense which do not in fact obtain. Similarly, if the necessity were natural, then the 'grounds' of this necessity would be found in nature, and Kant explicitly denies this. The necessity is in fact imperative, the kind of necessity that may be logically derived from sentences containing 'ought'. Kant captures a feature of ordinary language use when he says 'ought' expresses this necessity, because we often say 'it is imperative you do X ', 'it is essential you do X ' (1) is a claim that the reasons why moral imperative are true cannot be empirical. This clears the way for them to be purely rational. (2) must be read to mean when a person wills then what that person wills has no empirical cause. This is part of the doctrine of practical freedom, and need not be taken to preclude the possibility of akrasia. (3) divides into two. When Kant says ought expresses a possible action he means there are embedded in moral imperatives descriptions of actions that may be performed, so that in sentences of the form 'A ought to ', '' designates some performance that A is able to do. The second part of (3) is that the 'ground' of this possible action is a 'mere concept'. There are two ways of taking 'ground' here, both of which Kant intends. 'Grounds' can mean 'non-empirical cause' and 'ground' can mean 'reason for'. The inference he wants, from (1), is that the non-empirical cause and the reason for an action may be identical with a 'mere concept'. If (1) and (2) are true then it is true that the 'ground' of a free action is not empirical. But does it follow from that that it is merely conceptual? This depends whether 'empirical' and 'merely conceptual' designate mutually exhaustive categories, and, in particular, whether if X is not empirical then X is merely conceptual. It at least follows that if X is not empirical then it is possible that X is merely conceptual and in fact this all Kant needs for his purpose in (3) if we assume that if X is a reason (as well as a

Stephen Priest

Page 48

05/12/2007

cause) then X must either be or contain some conceptual items. If ground just meant 'cause' then there is nothing to force the conclusion that the ground must be a concept, but as 'ground' also includes the sense of 'reason', it is difficult to deny that if X is a ground in that sense then X must make sense, and if X makes sense, X contains concepts. Then 'mere' goes though easily as 'non-empirical'. The concept Kant primarily has in mind is 'ought', but if the ground of a free action is a reason as well as a cause then it would be safer to make that reason a sentence: either an imperative or a moral indicative, so that meta-sentences about a persons free actions could be constructed with the form: 'He 's because he ought to '. Here 'He ought to ' is a moral imperative, but it needs to be embedded in an indicative sentence as a clause introduced by 'because' for some close synonym in order for it to count as a sentence about the reasons for a free action. If someone wants the 'ground' in the sense of reason for a free action, then any such inquiry will take the logical form of 'Why did he ?' and coherent reply to that question could not be a mere concept, nor a straightforward moral imperative, but it would have to be a certain concept 'ought' (or a close synonym) embedded in a moral imperative embedded in turn in an indicative sentences as the 'because' clause. We can allow Kant this necessary background to 'mere concept' because there could not be concepts without a semantic role, and this particular one is suggested by the logical of 'ground' as 'reason' and by ordinary language uses of 'ought'. (4) is a truth about the imperative variety of necessity. It expresses the claim that from the fact that 'X ought to happen', it does not follow that either 'X has happened' or 'X will happen'. Kant might well have added that it does not follow that 'X is happening' either, because that is also true. (4) is just a report about the logic of 'ought'. Kant derives three conclusions from these premises, two weak and one strong. The weak ones are (c') and 'we represent (reason) to ourselves as having causality' from (c''). The strong one is 'our reasons has causality', from (c''). The first two go though without much difficulty. if there are non-empirical causes for actions, and these are accurately reported by indicative sentences in which imperatives are embedded, and if practical reasons contains the logic of 'ought' then it is true that 'it is possible for reason to have causality', and we think or our reason as having causality. These conclusions follow from our ordinary use of moral language; the first from the coherence of that language, the second from its role in how we think of ourselves as moral agents. The difficulty lies with the strong conclusion. From the fact that the members of some set of sentences S[...] Sn are each internally coherent, and also mutually consistent it does not follow that any of them is true. Kant has yet to show that reports of moral imperatives are truth valued, and still further that a class of them take the value 'true'. Secondly, it has not been shown either what justifies our conception of ourselves as moral agents at all. Kant has claimed there are no empirical features descriptions of which might feature as premises of such a justification. But a great deal of further argument is required to show that the language or morals has a genuine application. To say that we need it to make sense

Stephen Priest

Page 49

05/12/2007

of our moral behaviour is to be beg the question. There might be some alternative account of that behaviour under which it did not turn out to be moral in Kant's sense, or even moral at all. Reason and Freedom There are two further important facts about ourselves as moral agents with which Kant wishes to acquaint us. These are: reason is not deterministic, and reason is not determined. He has an argument for each of these. This is the one designed to convince us that we still perform our actions freely, even though they are caused by reason' (a) The Argument for the Freedom of Reason (1) '(the) empirical character (of reason) is itself determined in the intelligible character.' [da sie [die Vernunft] im empirschen Charakter derselben (der Sinnesart) ganz genau bestimmt und notwendig ist. Dieser ist wiederrum im intelligibelen Charakter bestimmt]. (CPR 475, A551, B579) (2) 'The action, in so far as it can be ascribed to a mode of thought does not follow there from in accordance with empirical laws.' [Die Handlung nun, in so fern sie der Denkungsart, als ihrer Ursache, beizumessen ist, erfolgt dennoch daraus nicht nach empirischen Gesetzen. ] (3) 'It is not preceded by the conditions of pure reason but only by their effects in the (field of) appearance of inner sense.' [ D.i. [nicht] so, da die Bedingungen der reinen Vernunft, sondern nur so, da deren Wirkungen in der Erscheinung des inneren Sinnes vorhergehen.] (4) 'The causality of reason in its intelligible character does not, in producing an effect, arise or begin to be at a certain time. For in that case [[...]] its causality would be nature not freedom.' [Die Kausalitt der Vernunft im intelligibelen Charakter entsteht nicht, oder hebt etwa zu einer gewissen Zeit an, um eine Wirkung hervorzubringen.] (CPR 475, A551, B579-80)

Stephen Priest

Page 50

05/12/2007

(5) 'If reason can have causality in respect of appearances, it is a faculty through which the sensible conditions of an empirical series of effects first begin.' [Wenn Vernunft Kausalitt in Ansehung der Erscheinungen haben kann; so ist sie ein Vermgen, durch welches die sinnliche Bedingung eienr empirischen Reieh von Wirkungen zuerst anfngt] (CPR 475, A552, B580) (c) 'Granted, then, that reason may be asserted to have causality in respect of appearance, its action can still be said to be free.' [Gesetzt nun, man knnte sagen: die Vernunft habe Kausalitt in Ansehung der Erscheinung; knnte da wohl die Handlung derselben freu heien.] (CPR 475, A551, B579) Premise (1) is intended to or stall the objection to reason's freedom that it may be empirically determined and hence itself be determining of its effects. Far from this being the case, Kant wants to claim, pure (non-empirical) reason itself determines its own 'empirical character'. There is prima facie a gross inconsistency in Kant's claiming that pure practical reason has any empirical dimension at all, but this inconsistency is an illusion. This is because the claim is merely that there exists empirical evidence for the operations of pure practical reason, and this is not at all to claim that what is utterly non-empirical is itself is also empirical: '[[...]] the will of every man has an empirical character, which is nothing but a certain causality of his reason, so far as that causality exhibits in its effects in the (field of) appearance, a rule form which we may gather what in their kind and degrees are the actions of reason, and the grounds therefore.' [So hat denn jeder Mensch einen empirischen Charakter seiner Willkr, welcher nichts anders ist, als eine gewisse Kausalitt seiner Vernunft, so fern diese an ihren Wirkungen in der Erscheinung eine Regel zeigt, darnach man die Vernunftgrnde und die Handlungen derselben nach ihrer Art und ihren Graden abnehmen, und die subjektiven Prinzipien seiner Willkr beurteilen kann] (CPR 474, A449, B577) So, the only causal connection thus asserted is between the operations of pure practical reason and its effects. (2) entails that this causal relation is not itself an empirical one. This accords with the doctrine of transcendental freedom because if it is true that a non-empirical cause has an empirical effect it is not clear that we should call the resulting causal connection itself 'empirical'. This is Kant's view and his ground of writ is that that causal link cannot be subsumed under any empirical generalisations, for example, it

Stephen Priest

Page 51

05/12/2007

cannot be captured by natural law. That is the sense in which reason's effects do not follow from it. There remain two residual sense in which the effect does clearly follow though. Firstly, the empirical event is in some sense still caused by reason. Secondly, although we cannot say that reason's causal efficacy is followed by an empirical effect, we may say that the agents decisions motives and reasons are formulated antecedently (in time) to the occurrence of what is in fact an empirical effect of pure-practical reason which is the claim spelled out in (3). Here the effects which appear to inner sense I interpret as precisely the introspectively available motive, intentions and reasons of the agent in performing an action. These are themselves further effects of pure practical reason and not the real causes of the agent's behaviour if he acts freely. So, although empirical motives, and decision for an action precede that action in time, pure practical reason does not precede it. This is not duplicated by but is confirmed by (4) which is the claim that purepractical reason, despite its causal efficacy is not itself in time. This is the force of its not arising or beginning to arise at a certain time. We may leave aside the question of whether it makes sense in Kantian terms to speak of pure practical reason lasting (externally). Here he wishes just to establish that pure practical reason is not an event or a happening that may be available to -for example-inner sense as decision might be held to be. Kant rightly notes that if reason did possess causal efficacy at some specifiable time 't' then it would have to be located within nature. This follows from one conclusion of the Transcendental Aesthetic; viz that there is only one time and putatively distinct times are in fact parts of one and the same time. (5) is an attempt to ex plain the sort of causal efficacy reason has over empirical events. 'Through' is crucially ambivalent but this much can be said about it. Reason may initiate empirical causal chains; it is 'through' it that an 'empirical series of events first begins'. Kant needs to give 'through' a content, and a theory of causal efficacy is just missing here, but the outline of one may be reconstructed. Two clues to this theory may be gleaned from the function it must perform in the theory of freedom in general. Firstly, the causation must not be empirical, and thus it must not be the sort of causation that obtains in the empirical world. Secondly, the theory must leave room for akrasia, that is for a person's actions to be empirically determined against his will even if in principle he could have done other than what he was empirically caused to do. It seems to me that the only realistic option open to Kant here is a probabilistic one; that the exercise of my freedom makes more likely the occurrence of a particular empirical event than its non-exercise otherwise would have done. If we try to analyse the efficacy of the rational in terms of necessary conditions, then that precludes any sort of action performed freely being empirically determined. If we interpret this causality in terms of sufficient conditions its efficacy becomes deterministic and the exercise of the will could never fail to be successful, no matter what the empirical pressure on the agent to act to the contrary. Kant does not anywhere so far as I can see subscribe to a probabilistic account of the

Stephen Priest

Page 52

05/12/2007

causal efficacy of the rational will, but it is the only one that convincingly fills the lacuna left by premise (5). Assuming (1)-(5) does (c) follow? In (c) I interpret 'it's action' as 'action caused by reason' and not as 'operation of reason itself'. If 'free' means 'not empirically determined' then that actions which are caused by pure practical reason are free does follow from the premises. If 'free' means 'uncaused' then the conclusion does not go through because a fortiori if my actions are caused by pure-practical reason they are caused. If we ask 'Am I free?' and if this depends upon whether my actions are free, then 'I am free' is true on the first interpretation' but false on the second: unless that is - as Kant will wish to argue - I am free if I act from pure practical reason. Clearly also 'its action' could be read to mean 'reason's own operations', in which case a shift has been made to the question of whether reason is itself determined, and to that we should now turn. (b) The Argument that Reason is Not Determined (1) 'Reason is not itself an appearance, and it is not subject to any conditions of sensibility.' [Vernunft [ist] selbst keine Erscheinung und gar keinen Bedingungen der Sinnlichkeit unterworfen []] (CPR 476, A553, B581) (2) 'The dynamical law of nature, which determines succession in time in accordance with rules is not applicable to it [reason]. [Auf sie [d.i. die Vernunft] kann also das dynamische Gesetz der Natur, was die Zeitfolge nach Regeln bestimmt, nicht angewandt werden.]' (3) 'Even as regards its causality there is no[t] time sequence in it.' [So findet in ihr [d.i die Vernunft] selbst in Betreff ihrer Kausalitt, keine Zeitfolge statt.] (4) 'Of pure reason we cannot say that the state wherein the will is determined is preceded and itself determined by some other state.' [Aber von der Vernunft kann man nicht sagen, da vor demjenigen Zustande, darin sie die Willkr bestimmt, ein anderer vorhergehe, darin dieser Zustand selbst bestimmt wird.] (c) 'Reason therefore acts freely; it is not dynamically determined in the chain of natural causes through either outer or inner grounds antecedent in time.'

Stephen Priest

Page 53

05/12/2007

[D[ie] reine [] Vernunft, welche mithin frei handelt, ohne in der Kette der Naturursachen, durch uere oder innere, aber der Zeit nach vorhergehende Grnde, dynmamisch bestimmt zu sein.] The first conjunct of (1) is analytic, but the second is question-begging unless we may assume that that which is itself non-empirical is not subject to empirical causes. The trouble with accepting this is that Kant importantly allows that empirical events may have non-empirical causes. If there may be empirical effects of noumenal and rational causes as the theory of freedom requires a special argument is needed to show that the converse is not equally permissible; that there may be noumenal and purely rational effects of empirical causes. This is a conclusion that would have appalled Kant, but one that he fails to pre-empt. (2), unpacked, is the claim that reason is non-temporal, natural laws are applicable only to what is temporal, so natural laws are not applicable to reason. That is unproblematic. (3) follows from (2) if reason's causality is a part of what reason is, and there is no difficulty by 'some other state'. That would seem not to follow if, in principle there could exist non-empirical, and non-temporal determinants of reason: God for example. From the fact that reason is not empirically determined it does not follow that reason is not non-empirically determined if there exist non-empirical determinants. If there are no non-empirical determinants of reason then (c) goes through: if it is true that reason is not empirically determined then it follows that reason is free. This is because of the meaning Kant gives 'free' in (c): 'not dynamically determined in the chain of natural causes'. Here natural causes are empirical causes. The point about inner and outer in (c) is meant to preclude a person, if he acts freely, being empirically determined to do so either by his own inclinations, motives etc (which are 'inner') or by some stimulus which is external to him (and so 'outer'). We may accept the inner-outer distinction here as corresponding to 'what may be detected by inner and outer sense' respectively. (7)

Stephen Priest

Page 54

05/12/2007

Notes (1) Bennett comments on (B 474) [] the phrase is not sufficient cause as both Schopenhauer and Kemp Smith imply, but rather cause [which is] sufficiently determined [a priori]. I do not know what that means. (Bennett, 1974: 185) Both Bennett and Kemp Smith allude to Schopenhauer The World as Will and Representation trans. E. F. J. Payne (New York, 1958) pp. 497-8 (Kemp Smith, 1979: 493) Kant means a cause which could in principle be discerned as the type of cause it is. It is true a priori of such a cause that it is in principle discoverable because empirical. (It is legitimate to translate bestimmt as determined but determined (like bestimmt) is ambiguous between, on the one hand, causally determined (causally necessitated) and, roughly, discriminated or made out (as the thing or sort of thing it is). We need determined in the second sense to make sense of cause [which is] sufficiently determined [a priori]. Bennett is right that Schopenhauer and Kemp Smith have not read Kant accurately at this point but does understand what they have omitted. By cause Kant still means sufficient condition here (even though that is not the meaning of the controversial phrase). A cause is contrasted with a state of the world which in nowise follows from an earlier state in the Antithesis-argument. If we read Schopenhauer and Kemp Smith as right and Bennett as wrong then Thesis and Antithesis do come out as clearly mutually inconsistent, as Kant intends them. We do not have to go with their misreading because in the Antinomies Kant thinks of causes as sufficient conditions anyway. By going so far as to reject cause as is a sufficient condition for Bennett is forced to conclude: I confess defeat: I cannot discover how the Thesis-argument of the third antinomy is supposed to work. (Bennett, 1974: 187) (2) Bennett says: According to [Lewis-White] Beck, the Thesis-argument is largely a repetition of the Aristotelian-Thomistic proof of the impossibility of an infinite series of causes and hence of the necessity of a first cause. (Bennett, 1974: 187) There is a mistake here. Both Aristotle and Aquinas think there is no sound refutation of the claim that the regress of past events might be infinite. By prime in prime mover Aristotle means necessary condition that has no necessary condition except itself and Aquinas thinks that an infinite God is needed to everlastingly sustain a universe that cannot be proven to have begun. That there was a first event (as implied by Genesis) is a matter of faith, not proof, according to Aquinas. Indeed, Aquinas thinks he has a proof that there cannot be such a proof.

Stephen Priest

Page 55

05/12/2007

(3) Bennett rightly reports Kants view that there is another kind of causality [than natural causality] -that of freedom- in which an events cause does not precede the event itself (Bennett, 1974: 187) but does not report Kants point that it is not simultaneous with it either. The causality of freedom is noumenal causality and any putatively noumenal causality is atemporal causality and so not simultaneous with anything. Of course Kant, in his better moments, thinks this barely coherent given that causation is a category and the categories only have empirical uses. (Bennetts footnote ascribes Kants solution to the Third Antinomy to [A] 560-86 but it is at A 460-86. This is no doubt a typographical slip. (Bennett, 1974: 187 fn 10)) (4) The answer to Bennetts question If the concept of freedom solves problems about humans why invoke cosmology in its defence? (Bennett, 1974: 189) is that one putative cosmology entails that there was an uncaused cause (because a first cause that was not cause of itself). Of course, Kant thinks the problem of the origin of the universe insoluble (because it presupposes the erroneous doctrine of transcendental realism) but uncaused cause might be exhibited as free from contradiction even if devoid of cosmological reference. It might then be given at least prima facie legitimate application in analyzing free. Bennett is too hard on Kant when he writes [] the alignment of Thesis/Antithesis with practical/theoretical is totally baseless. (Bennett, 1974: 189) The freedom entailed by the conclusion of the Thesis is necessary for morality, and so practical reasoning. By theoretical Kant means reason in its applications to the empirical world, which, according to the conclusion of the Antithesis, is deterministic. (5) Bennett says: If Kant is right that thinking involves concepts and that these are tools for the orderly management of intuitions, then the theory of noumenal freedom does not even make sense. (Bennett, 1974: 194) Bennett needs essentially after thinking and only before tools, then his claim goes through. (I assume p does not make sense here if p cannot be true or false.) (6) Kemp Smith is right to argue that What [] is to be said regarding the truth of the antithesis? If the principle of natural causality be formulated as asserting that every event has an antecedent cause determining it to exist, then certainly, free, spontaneous and self-originating causality is impossible (Kemp Smith, 1933: 494) if we read antecedent to mean chronologically antecedent. His claim becomes doubtful if we read antecedent to mean prior in the sense of necessary. That some cause is a necessary condition for some effect does not per se entail that it is chronologically antecedent, even if we supply the assumption that all causes are events. Kemp Smith is right to say this antinomy is chiefly concerned with the problem of freedom, i.e. of spontaneous origination within the world. (Kemp Smith,

Stephen Priest

Page 56

05/12/2007

1933: 495) Kant is trying to explain how our actions can possibly make a difference within the empirical causal order. (7) The account of Kant on freedom I have offered provides a way of overcoming the dispute between Allison and Guyer on incorporation. (Allison, 1990: 50 ff., Guyer, 1992: 102 ff.) Allison thinks practical freedom presupposes transcendental freedom. There is some intuitive plausibility in Allisons claim that: [incentives] do not motivate by themselves causing action but rather by being taken as reasons and incorporated into maxims. (Allison, 1990: 51) partly because that a has some incentive to does not look like a sufficient condition for as -ing. We sometimes fail to act despite the existence of incentives to act and, sometimes, despite our knowledge of those incentives to act. What does Allison mean by incorporation? He provides us with a necessary but not a sufficient condition for some acts being an act of incorporation when he says: [An] act of incorporation as an expression of spontaneity, cannot be regarded as a causal consequence of antecedent conditions. (Allison, 1990: 52) Guyer says But this is clearly false (Guyer, 1992: 105) Guyers reasons are: [1] the fact that a particular agent holds a certain principle and has justifed or excluded action on a given incentive on the basis of holding this principle is just as temporal a fact as the fact that a certain desire occurs in the agent. Holding a principle may be a disposition rather than an event, but it is still temporal. (Guyer, 1992: 105) [2] applying or deciding to apply a certain principle to a certain desire may be as much of a temporal event as the occurrence of the desire itself. (Guyer, 1992: 105) [3] the incorporation of [] incentives into our maxims or principles hardly removes our rational agency from the sphere of temporal discourse by itself. (Guyer, 1992: 105) I fully agree with Guyer that all that is temporal. However Guyer thinks the choice we are left with is exhaustive: We must either accept this result, whatever its implications for freedom, or defend the unpleasant doctrine of timeless events of noumenal choice. (Guyer, 1992: 105) (I take it this is unpleasant because timeless event looks contradictory.)

Stephen Priest

Page 57

05/12/2007

I suggest there is a third possibility. Kant, like any coherent libertarian, is committed to the thesis that if I act freely I might have not acted. The possibility of not acting is not something temporal. I suggest it is this feature that Kant thinks is presupposed by actions (rather than mere events). It is a timeless modal property, not any kind of phenomenal or empirical property. With this assumption spelled out, Allisons view goes through, immune to Guyers objections. Practical freedom presupposes transcendental freedom. Kant Allison, H. (1983) Kant's Transcendental Idealism: An Interpretation and Defense. New Haven, CN: Yale University Press. Allison, H.E. (1990) Kant's Theory of Freedom, Cambridge: Cambridge University Press. Allison, H.E. (1996) Idealism and Freedom: Essays on Kant's Theoretical and Practical Philosophy, Cambridge: Cambridge University Press. Ameriks, K. 2000. Kant's Theory of Mind: An Analysis of the Paralogisms of Pure Reason, 2nd edition.Ox ford: Ox ford University Press. Beck, L.W. (1965) Studies in the Philosophy of Kant, Indianapolis, IN: BobbsMerrill Company. Beck, L.W. (1969) Early German Philosophy: Kant and his Predecessors, Cambridge, MA: Harvard University Press. Beck, L.W. (1978) Essays on Kant and Hume, New Haven and London: Yale University Press. Beiser, F.C. (1987) The Fate of Reason: German Philosophy from Kant to Fichte, Cambridge, MA: Harvard University Press. Bennett, J. (1974) Kant's Dialectic. Cambridge: Cambridge University Press. Brook, A. (1993) Kant's A Priori Methods for Recognizing Necessary Truths In Return of the A Priori, Philip Hanson and Bruce Hunter, eds. Canadian Journal of Philosophy, Supplementary Volume 18, pp. 215-52. Brook, A. (1994) Kant and the Mind. Cambridge and New York: Cambridge University Press.

Stephen Priest

Page 58

05/12/2007

Brook, A. (1998) Critical Notice of L. Falkenstein, Kant's Intuitionism: A Commentary on the Transcendental Aesthetic. Canadian Journal of Philosophy 29, pp. 247-68. Brook, A. (2001) Kant on self-reference and self-awareness In A. Brook and R. DeVidi, eds. 2001. Brook, A. (2004) Kant, cognitive science, and contemporary neo-Kantianism. In D. Zahavi, ed. Journal of Consciousness Studies, special number. Caygill, H. (1995) A Kant Dictionary, Ox ford: Blackwell. Chadwick, R. (ed.) (1992) Immanuel Kant: Critical Assessments, London: Routledge, 4 vols. Dryer, D.P. (1966) Kant's Solution for Verification in Metaphysics, London: George Allen & Unwin. Falkenstein, L. (1995) Kant's Intuitionism: A Commentary on the Transcendental Aesthetic. Toronto: University of Toronto Press. Frster, E. (ed.) (1989) Kant's Transcendental Deductions: The Three 'Critiques' and the 'Opus postumum', Stanford: Stanford University Press. Gardner, S. (1999) Kant and the Critique of Pure Reason , London: Routledge. Gregor, M. (1963) Laws of Freedom: A Study of Kant's Method of Applying the Categorical Imperative in the Metaphysik der Sitten, Ox ford: Basil Blackwell. Grier, M. (2001) Kant's Doctrine of Transcendental Illusion, Cambridge: Cambridge University Press. Guyer, P. (1993) Kant and the Experience of Freedom: Essays on Aesthetics and Morality, Cambridge: Cambridge University Press. (A collection of essays placing Kant's aesthetics in its historical contex t and ex ploring connections with Kant's moral philosophy.) Guyer, P. (2000) Kant on Freedom, Law, and Happiness, Cambridge: Cambridge University Press. (Essays on the development of Kant's moral philosophy and on its foundations, applications in political philosophy and ethics, and philosophy of history.) Guyer, P. (1987) Kant and the Claims of Knowledge. Cambridge and New York: Cambridge University Press.

Stephen Priest

Page 59

05/12/2007

Guyer, P. (ed.) (1992) The Cambridge Companion to Kant, Cambridge: Cambridge University Press. Guyer, P. (ed.) (1998) Kant's Groundwork of the Metaphysics of Morals: Critical Essays, Lanham, MD: Rowman & Littlefield. Henrich, D. (1976) Identitt und Objektivitt. Heidelberg: Carl Winter UniversittsVerlag. Henrich, D. (1994) The Unity of Reason: Essays on Kant's Philosophy, ed. R. Velkley, Cambridge, MA: Harvard University Press. Hffe, O. (1994) Immanuel Kant, trans. M. Farrier, Albany: State University of New York Press. Howell, R. (1992) Kant's Transcendental Deduction: An Analysis of Main Themes in his Critical Philosophy, Dordrecht and Boston: Kluwer. Kemp Smith, N. (1958) A Commentary to Kants Critique of Pure Reason. London: Macmillan. Kitcher, P. (1990) Kant's Transcendental Psychology. New York: Ox ford University Press. Klemme, H.F. and Kuehn, M (eds) (1999) Immanuel Kant, Ashgate: Dartmouth, 2 vols. Korner, S. (1955) Kant. Harmondsworth: Penguin Books. Kuehn, M. 2001. Kant: A Biography. Cambridge and New York: Cambridge University Press. Meerbote, R. 1989. Kant's functionalism. In: J. C. Smith, ed. Historical Foundations of Cognitive Science. Dordrecht, Holland: Reidel. Paton, H.J. (1936) Kant's Metaphysics of Experience: A Commentary on the First Half of the Kritik der reinen Vernunft, London: George Allen & Unwin, 2 vols. Pippin, R. (1987) Kant on the spontaneity of mind. Canadian Journal of Philosophy 17, pp. 449-476. Prauss, G. (1974) Kant und das Problem der Dinge an sich (Kant and the Problem of the Ding an sich) , Bonn: Bouvier.

Stephen Priest

Page 60

05/12/2007

Sassen, B. (2000) Kant's Early Critics. Cambridge and New York: Cambridge University Press. Sellars, W. (1970) this I or he or it (the thing) which thinks . Proceedings of the American Philosophical Association 44, pp. 5-31. Strawson, P. F. (1966) The Bounds of Sense. London: Methuen. Vaihinger, H. (1881-92) Commentar zu Kants Kritik der reinen Vernunft (Commentary on Kant's Critique of Pure Reason) , Stuttgart: W. Spemann and Union Deutsche Verlagsgesellschaft, 2 vols. Van Cleve, J. (1999) Problems from Kant, New York: Ox ford University Press. De Vleeschauwer, H.J. (1934-7) La Dduction Transcendentale dans l'vre de Kant (The Transcendental Deduction in the Work of Kant) , Antwerp, Paris, and the Hague: De Sikkel, Champion, and Martinus Nijhoff, 3 vols. De Vleeschauwer, H.J. (1962) The Development of Kantian Thought: The History of a Doctrine, trans. A.R.C. Duncan, London: Thomas Nelson. Walker, R. C. S. (1978) Kant. London: Routledge, Kegan Paul. Ward, K. (1972) The Development of Kants Ethics. Ox ford: Blackwell. Waxman, W. (1991) Kant's Model of the Mind. New York: Ox ford University Press. Wolff, R.P. (ed.) (1967) Kant: A Collection of Critical Essays, Garden City: Doubleday Anchor. Wolff, R.P. (1963) Kant's Theory of Mental Activity: A Commentary on the Transcendental Analytic of the Critique of Pure Reason, Cambridge, MA: Harvard University Press. Wood, A.W. (1984) Self and Nature im Kants Philosophy (ed.), Ithaca, Cornell. Wood, A.W. (1984a) Kants Compatibilism in Wood (1984) [NB: Move to Kants Attack on Rational Psychology. That there is free will is part of Rational Psychology (endorsed by Plato, Descartes, Christian philosophy, a priori metaphysics) but not repudiated by Kant (and possibly not recognized by him, as such, to be part of Rational Psychology). Is Kants compatibilism logically defensible?]

Stephen Priest

Page 61

05/12/2007

[NB: Freedom is spared in Kants attack on Rational Psychology. Is he entitled to spare freedom? His arguments for freedom should be the model for arguing about the soul.] [NB: Move to Kants Attack on Rational Psychology when replacable. 4th Paralogism is part of Rational Psychology. Refutation of Idealism is a continuation of Kants attack on Rational Psychcology.] Freedom Adams, Robert (1987). Must God Create the Best?, in The Virtue of Faith and Other Essays in Philosophical Theology. New York: Ox ford University Press, 5164. Aquinas, Thomas (1945). Basic Writings of Saint Thomas Aquinas (2 vols.). New York: Random House. ----- (1993). Selected Philosophical Writings, ed. T. McDermott. Ox ford: Ox ford University Press. Aristotle (1985). Nichomachean Ethics, translated by Terence Irwin. Indianapolis: Hackett Publishing. Augustine (1993). On the Free Choice of the Will, tr. Thomas Williams. Indianapolis: Hackett Publishing. Ayer, A.J. (1982). Freedom and Necessity, in Watson (1982b), ed., 15-23. Baker, Lynne (2000). Persons and Bodies: A Constitution View. Cambridge: Cambridge University Press. Bourke, Vernon (1964). Will in Western Thought. New York: Sheed and Ward. Campbell, C.A. (1967). In Defence of Free Will & other essays. London: Allen & Unwin Ltd. Chisholm, Roderick (1982). Human Freedom and the Self, in Watson (1982b), 2435. ----- (1976). Person and Object. LaSalle: Open Court. Clarke, Randolph (1993). Toward a Credible Agent-Causal Account of Free Will, in O'Connor (1995), ed., 201-15.

Stephen Priest

Page 62

05/12/2007

----- (1995). Indeterminism and Control, American Philosophical Quarterly 32, 125-138. ----- (1996). Agent Causation and Event Causation in the Production of Free Action, Philosophical Topics 24 (Fall), 19-48. ----- (2003). Libertarian Accounts of Free Will. Ox ford: Ox ford University Press. Dennett, Daniel (1984). Elbow Room: The Varieties of Free Will Worth Having. Cambridge. MA: MIT Press. Descartes, Ren (1984). Meditations on First Philosophy [1641] and Passions of the Soul [1649], in The Philosophical Writings of Descartes, vol. I-III, translated by Cottingham, J., Stoothoff, R., & Murdoch, D.. Cambridge: Cambridge University Press. Donagan, Alan (1985). Human Ends and Human Actions: An Ex ploration in St. Thomas's Treatment. Milwaukee: Marquette University Press. Dilman, Ilham (1999). Free Will: An Historical and Philosophical Introduction. London: Routledge. Double, Richard (1991). The Non-Reality of Free Will. New York: Ox ford University Press. Edwards, Jonathan (1957) [1754]. Freedom of Will, ed. P. Ramsey. New Haven: Yale University Press. Ekstrom, Laura (2000). Free Will: A Philosophical Study. Boulder, CO: Westview Press. Farrer, Austin (1958). The Freedom of the Will. London: Adam & Charles Black. Fischer, John Martin (1994). The Metaphysics of Free Will. Ox ford: Blackwell. ----- (1999). Recent Work on Moral Responsibility, Ethics 110, 93-139. Fischer, John Martin and Ravizza, Mark. (1992). When the Will is Free, in O'Connor (1995), ed., 239-269. Fischer, John Martin (1998) Responsibility and Control. Cambridge: Cambridge University Press. Frankfurt, Harry (1969). Alternate Possibilities and Moral Responsibility, Journal of Philosophy 66, 829-39.

Stephen Priest

Page 63

05/12/2007

----- (1982). Freedom of the Will and the Concept of a Person, in Watson (1982), ed., 81-95. ----- (1988). The Importance of What We Care About. Cambridge: Cambridge University Press. ----- (1992). The Faintest Passion, Proceedings and Addresses of the American Philosophical Association 66, 5-16. Ginet, Carl (1990). On Action. Cambridge: Cambridge University Press. ----- (1997). Freedom, Responsibility, and Agency, The Journal of Ethics 1, 85-98. ----- (2002) Reasons Ex planations of Action: Causalist versus Noncausalist Accounts, in Kane, ed., (2002), 386-405. Hobbes, Thomas and Bramhall, John (1999) [1655-1658]. Hobbes and Bramhall on Liberty and Necessity, ed. V. Chappell. Cambridge: Cambridge University Press. Honderich, Ted (1988). A Theory of Determinism. Ox ford: Ox ford University Press. Howard-Snyder, Daniel and Frances (1994). How an Unsurpassable Being Can Create a Surpassable World, Faith and Philosophy 11, 260-8. Howard-Snyder, Daniel and Moser, Paul, eds. (2002). Divine Hiddenness: New Essays. Cambridge: Cambridge University Press. Hume, David (1977) [1748]. An Enquiry Concerning Human Understanding. Indianapolis: Hackett Publishing. Kane, Robert (1995). Two Kinds of Incompatibilism, in O'Connor (1995), ed., 115150. ----- (1996). The Significance of Free Will. New York: Ox ford University Press. Kane, Robert, ed., (2002). Ox ford Handbook on Free Will. New York: Ox ford University Press. Kant, Immanuel (1993) [1788]. Critique of Practical Reason, tr. by Lewis White Beck. Upper Saddle River, NJ: Prentice-Hall Inc. Kretzmann, Norman (1997). The Metaphysics of Theism: Aquinas's Natural Theology in Summa Contra Gentiles I. Ox ford: Clarendon Press. Leibniz, Gottfried (1985) [1710]. Theodicy. LaSalle, IL: Open Court.

Stephen Priest

Page 64

05/12/2007

Libet, Benjamin (2002). Do We Have Free Will? in Kane, ed., (2002), 551-564. MacDonald, Scott (1998). Aquinas's Libertarian Account of Free Will, Revue Internationale de Philosophie, 2, 309-328. Magill, Kevin (1997). Freedom and Experience. London: MacMillan. McCann, Hugh (1998). The Works of Agency: On Human Action, Will, and Freedom. Ithaca: Cornell University Press. Mele, Alfred (1995). Autonomous Agents (New York: Ox ford University Press). Morris, Thomas (1993). Perfection and Creation, in E. Stump. (1993), ed., 234-47 Murray, Michael (1993). Coercion and the Hiddenness of God, American Philosophical Quarterly 30, 27-38. ----- (2002). Deus Absconditus, in Howard-Snyder amd Moser (2002), 62-82. Nozick, Robert (1995). Choice and Indeterminism, in O'Connor (1995), ed., 10114. O'Connor, Timothy (1993). Indeterminism and Free Agency: Three Recent Views, Philosophy and Phenomenological Research, 53, 499-526. -----, ed., (1995). Agents, Causes, and Events: Essays on Indeterminism and Free Will. New York: Ox ford University Press. ----- (2000). Persons and Causes: The Metaphysics of Free Will. New York: Ox ford University Press. ----- (forthcoming). Freedom With a Human Face, Midwest Studies in Philosophy, Fall 2005. Pettit, Philip (2001). A Theory of Freedom. Ox ford: Ox ford University Press. Pereboom, Derk (2001). Living Without Free Will. Cambridge: Cambridge University Press. -----, ed., (1997). Free Will. Indianapolis: Hackett Publishing. Pink, Thomas (2004). Free Will: A Very Short Introduction. Ox ford: Ox ford University Press. Plato (1997). Complete Works, ed. J. Cooper. Indianapolis: Hackett Publishing.

Stephen Priest

Page 65

05/12/2007

Quinn, Phillip (1983). Divine Conservation, Continuous Creation, and Human Action, in A. Freddoso, ed. The Existence and Nature of God. Notre Dame: Notre Dame University Press. Reid, Thomas (1969). Essays on the Active Powers of the Human Mind, ed. B. Brody. Cambridge: MIT Press. Rowe, William (1993). The Problem of Divine Perfection and Freedom, in E. Stump (1993), ed., 223-33. ----- (1995). Two Concepts of Freedom, in O'Connor (1995), ed. 151-71. ----- (2004). Can God Be Free?. Ox ford: Ox ford University Press. Sartre, John Paul (1956). Being and Nothingness. New York: Washington Square Press. Schopenhauer, Arthur (1999) [1839]. Prize Essay on the Freedom of the Will, ed. G. Zoller. Cambridge: Cambridge University Press. Scotus, John Duns (1986). Questions on Aristotle's Metaphysics IX , Q.15 in Duns Scotus on the Will and Morality [selected and translated by Allan B. Wolter, O.F.M.]. Washington: Catholic University of America Press. ----- (1994) [1297-99]. Contingency and Freedom: Lectura I 39, tr. Vos Jaczn et al. Dordrecht: Kluwer Academic Publishers. Shatz, David (1986). Free Will and the Structure of Motivation, Midwest Studies in Philosophy 10, 451-482. Smilansky, Saul (2000). Free Will and Illusion. Ox ford: Ox ford University Press. Strawson, Galen (1986). Freedom and Belief. Ox ford: Clarendon Press. Strawson, Peter (1982). Freedom and Resentment, in Watson (1982), ed., 59-80. Stump, Eleonore (1996). Persons: Identification and Freedom, Philosophical Topics 24, 183-214. -----(1997). Aquinas's Account of Freedom: Intellect and Will, The Monist 80, 576597. -----, ed., (1993). Reasoned Faith. Ithaca: Cornell University Press.

Stephen Priest

Page 66

05/12/2007

van Inwagen, Peter (1983). An Essay on Free Will. Ox ford: Ox ford University Press. ----- (1994). When the Will is Not Free, Philosophical Studies, 75, 95-113. ----- (1995). When Is the Will Free? in O'Connor (1995), ed., 219-238. Wainwright, William (1996). Jonathan Edwards, William Rowe, and the Necessity of Creation, in J. Jordan and D. Howard-Snyder, eds., Faith Freedom, and Rationality. Lanham: Rowman and Littlefield, 119-133. Wallace, R. Jay (2003). Addiction as Defect of the Will: Some Philosophical Reflections, in Watson, ed., (2003b), 424-452. Watson, Gary (1987). Free Action and Free Will, Mind 96, 145-72. ----- (2003a). Free Agency, in Watson, ed., 1982b. -----, ed., (2003b). Free Will. 2nd ed. Oxford: Oxford University Press. Wegner, Daniel (2002). The Illusion of Conscious Will. Cambridge, MA: MIT Press. Widerker, David and McKenna, Michael, eds., (2002). Moral Responsibility and Alternative Possibilities. Aldershot: Ashgate Publishing. Wolf, Susan (1990). Freedom Within Reason. Oxford: Oxford University Press. (x ) (p) Kx p p (x ) (p) Kx p(t) p(t) (x ) () - & v (x ) (p) Kx p p (x ) (p) Kx p(t) p(t) (x ) () - & v